{"step": 1, "time": 1792444884.192339, "scalars": {"train/loss": -2.1420419216156006e-08, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.8405338525772095e-08, "train/kl": 0.0, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -2.1420419216156006e-08, "train/step_time_ms": 424.84116554260254, "train/samples_per_sec": 75.32226769769343, "train/data_wait_ms": 0.0, "train/accepted": 1.0, "train/grad_norm": 1.3693747520446777, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 5.047427895999988, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.0401785746216774, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 2, "time": 1792444884.2434864, "scalars": {"train/loss": 0.00039209285750985146, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 3.818422555923462e-08, "train/kl": 0.01960240863263607, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 4.470348358154297e-08, "train/step_time_ms": 91.90869331359863, "train/samples_per_sec": 348.17163476379596, "train/data_wait_ms": 0.062399999990248034, "train/accepted": 1.0, "train/grad_norm": 1.2721186876296997, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03734771799997816, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.0669642835855484, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 3, "time": 1792444884.2947528, "scalars": {"train/loss": 0.0009088036022149026, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.7474015951156616e-08, "train/kl": 0.04544055461883545, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -7.450580596923828e-09, "train/step_time_ms": 51.40995979309082, "train/samples_per_sec": 622.4474815539654, "train/data_wait_ms": 0.044380000019828, "train/accepted": 1.0, "train/grad_norm": 1.29212486743927, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.045955282000022635, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.0625, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 4, "time": 1792444884.3621404, "scalars": {"train/loss": 0.0016885431250557303, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.862645149230957e-09, "train/kl": 0.08442603796720505, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 2.2351741790771484e-08, "train/step_time_ms": 67.16322898864746, "train/samples_per_sec": 476.45118440343197, "train/data_wait_ms": 0.05364999998391795, "train/accepted": 1.0, "train/grad_norm": 1.334789514541626, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04732516299998224, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.0446428582072258, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 5, "time": 1792444884.408348, "scalars": {"train/loss": 0.002485061762854457, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.816079020500183e-08, "train/kl": 0.12425048649311066, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 5.21540641784668e-08, "train/step_time_ms": 46.11682891845703, "train/samples_per_sec": 693.8898608267676, "train/data_wait_ms": 0.04885000004151152, "train/accepted": 1.0, "train/grad_norm": 1.4557445049285889, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06164579900001854, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.1205357164144516, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 6, "time": 1792444884.4497979, "scalars": {"train/loss": 0.0026755183935165405, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 6.51925802230835e-09, "train/kl": 0.13377293944358826, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 5.960464477539063e-08, "train/step_time_ms": 41.79716110229492, "train/samples_per_sec": 765.6022360390166, "train/data_wait_ms": 0.060949999976855906, "train/accepted": 1.0, "train/grad_norm": 1.420271873474121, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03976995900001157, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.0758928582072258, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 7, "time": 1792444884.5132725, "scalars": {"train/loss": 0.006073155906051397, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -4.656612873077393e-10, "train/kl": 0.3036572337150574, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 1.1175870895385742e-08, "train/step_time_ms": 63.30227851867676, "train/samples_per_sec": 505.5110297579366, "train/data_wait_ms": 0.08550000001150693, "train/accepted": 1.0, "train/grad_norm": 1.3597381114959717, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.038000137999972594, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.1071428656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 8, "time": 1792444884.5595546, "scalars": {"train/loss": 0.004999002907425165, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.1420419216156006e-08, "train/kl": 0.24994902312755585, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 2.2351741790771484e-08, "train/step_time_ms": 45.89414596557617, "train/samples_per_sec": 697.2566833251946, "train/data_wait_ms": 0.047559999984514434, "train/accepted": 1.0, "train/grad_norm": 1.168060064315796, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.0596147190000238, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.1205357164144516, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 9, "time": 1792444884.6098511, "scalars": {"train/loss": 0.008103807456791401, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -4.237517714500427e-08, "train/kl": 0.40519002079963684, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 7.450580596923828e-09, "train/step_time_ms": 49.97396469116211, "train/samples_per_sec": 640.3334255698787, "train/data_wait_ms": 0.06723999996438579, "train/accepted": 1.0, "train/grad_norm": 1.4500458240509033, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04050614900000937, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.1160714328289032, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 10, "time": 1792444884.659335, "scalars": {"train/loss": 0.005594536196440458, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -6.51925802230835e-09, "train/kl": 0.2797276973724365, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.7578713595867157e-08, "train/step_time_ms": 50.299882888793945, "train/samples_per_sec": 636.184383783707, "train/data_wait_ms": 0.0470500000346874, "train/accepted": 1.0, "train/grad_norm": 1.300782322883606, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04368270100002292, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.0625, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 11, "time": 1792444884.7006485, "scalars": {"train/loss": 0.0059638298116624355, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.3969838619232178e-09, "train/kl": 0.298189640045166, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 3.725290298461914e-08, "train/step_time_ms": 41.2600040435791, "train/samples_per_sec": 775.5694828871412, "train/data_wait_ms": 0.10893000001033215, "train/accepted": 1.0, "train/grad_norm": 1.1106261014938354, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03394112600000199, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.1562500149011612, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 12, "time": 1792444884.766953, "scalars": {"train/loss": 0.006609700620174408, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.4435499906539917e-08, "train/kl": 0.3304835557937622, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 2.9802322387695312e-08, "train/step_time_ms": 66.32709503173828, "train/samples_per_sec": 482.45743288904225, "train/data_wait_ms": 0.044879999961722206, "train/accepted": 1.0, "train/grad_norm": 1.5641133785247803, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03656085800002984, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.1383928656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 13, "time": 1792444884.8137486, "scalars": {"train/loss": 0.011025954969227314, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.210719347000122e-08, "train/kl": 0.5512983202934265, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.1175870895385742e-08, "train/step_time_ms": 46.83566093444824, "train/samples_per_sec": 683.240064547986, "train/data_wait_ms": 0.14158999999835942, "train/accepted": 1.0, "train/grad_norm": 1.5472495555877686, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06084448900003281, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.2008928656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 14, "time": 1792444884.862595, "scalars": {"train/loss": 0.014562800526618958, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.1886080503463745e-08, "train/kl": 0.7281398773193359, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 3.725290298461914e-09, "train/step_time_ms": 48.80809783935547, "train/samples_per_sec": 655.6289102952383, "train/data_wait_ms": 0.0611199999980272, "train/accepted": 1.0, "train/grad_norm": 1.0962203741073608, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04249319000001606, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.2053571492433548, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 15, "time": 1792444884.914068, "scalars": {"train/loss": 0.010957470163702965, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.0710209608078003e-08, "train/kl": 0.5478711128234863, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 4.842877388000488e-08, "train/step_time_ms": 51.53465270996094, "train/samples_per_sec": 620.9414115992449, "train/data_wait_ms": 0.06595000002107554, "train/accepted": 1.0, "train/grad_norm": 1.6887049674987793, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04268246999998837, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.1473214328289032, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 16, "time": 1792444884.962652, "scalars": {"train/loss": 0.015524912625551224, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.561137080192566e-08, "train/kl": 0.776245653629303, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 48.39444160461426, "train/samples_per_sec": 661.2329626910894, "train/data_wait_ms": 0.12328099995784214, "train/accepted": 1.0, "train/grad_norm": 1.474214792251587, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04786370199997236, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.2544642984867096, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 17, "time": 1792444885.0122879, "scalars": {"train/loss": 0.01422944013029337, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -4.330649971961975e-08, "train/kl": 0.7114710807800293, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 1.862645149230957e-08, "train/step_time_ms": 49.58033561706543, "train/samples_per_sec": 645.4171719843235, "train/data_wait_ms": 0.04513999999744556, "train/accepted": 1.0, "train/grad_norm": 1.3972793817520142, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.042383379999989756, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.25, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 18, "time": 1792444885.0832148, "scalars": {"train/loss": 0.016829373314976692, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -4.6100467443466187e-08, "train/kl": 0.8414649963378906, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 7.450580596923828e-08, "train/step_time_ms": 71.0914134979248, "train/samples_per_sec": 450.12468349548425, "train/data_wait_ms": 0.04516000001331122, "train/accepted": 1.0, "train/grad_norm": 1.4461714029312134, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04293615099999215, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.2678571343421936, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 19, "time": 1792444885.1301377, "scalars": {"train/loss": 0.02181926928460598, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 5.3551048040390015e-08, "train/kl": 1.090965747833252, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -4.470348358154297e-08, "train/step_time_ms": 46.91433906555176, "train/samples_per_sec": 682.094230407627, "train/data_wait_ms": 0.06395000002612505, "train/accepted": 1.0, "train/grad_norm": 1.5794111490249634, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06641039200002297, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.2678571343421936, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 20, "time": 1792444885.1743429, "scalars": {"train/loss": 0.019937584176659584, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -5.448237061500549e-08, "train/kl": 0.9968775510787964, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 3.3527612686157227e-08, "train/step_time_ms": 44.2204475402832, "train/samples_per_sec": 723.647131134283, "train/data_wait_ms": 0.041310000028715876, "train/accepted": 1.0, "train/grad_norm": 1.2024434804916382, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.042215310000017325, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.2232142984867096, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 21, "time": 1792444885.2292237, "scalars": {"train/loss": 0.023683130741119385, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.6298145055770874e-08, "train/kl": 1.1841545104980469, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 4.0978193283081055e-08, "train/step_time_ms": 54.940223693847656, "train/samples_per_sec": 582.4512142200003, "train/data_wait_ms": 0.07205000002841189, "train/accepted": 1.0, "train/grad_norm": 1.2518565654754639, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03943319900002962, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.3080357313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 22, "time": 1792444885.2783306, "scalars": {"train/loss": 0.026081571355462074, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.8870999813079834e-08, "train/kl": 1.3040778636932373, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 1.4901161193847656e-08, "train/step_time_ms": 49.10135269165039, "train/samples_per_sec": 651.7132063744865, "train/data_wait_ms": 0.06815000000415239, "train/accepted": 1.0, "train/grad_norm": 1.206903100013733, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.05106427399999802, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.3705357313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 23, "time": 1792444885.3319943, "scalars": {"train/loss": 0.024748211726546288, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -4.656612873077393e-10, "train/kl": 1.2374109029769897, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -5.587935447692871e-09, "train/step_time_ms": 53.542375564575195, "train/samples_per_sec": 597.6574565954055, "train/data_wait_ms": 0.056870000037179125, "train/accepted": 1.0, "train/grad_norm": 1.6311925649642944, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.044540621000010105, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.3169642984867096, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 24, "time": 1792444885.3937204, "scalars": {"train/loss": 0.028743501752614975, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -7.916241884231567e-09, "train/kl": 1.4371764659881592, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -2.60770320892334e-08, "train/step_time_ms": 23.915529251098633, "train/samples_per_sec": 1338.042727970571, "train/data_wait_ms": 0.05447000000913249, "train/accepted": 1.0, "train/grad_norm": 1.3951290845870972, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.048523673000033796, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.4151785969734192, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 25, "time": 1792444885.4374452, "scalars": {"train/loss": 0.029753416776657104, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.3969838619232178e-09, "train/kl": 1.4876673221588135, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 7.078051567077637e-08, "train/step_time_ms": 81.67386054992676, "train/samples_per_sec": 391.8022214762162, "train/data_wait_ms": 0.048890000016399426, "train/accepted": 1.0, "train/grad_norm": 1.1611741781234741, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03455566700000645, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.4464285671710968, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 26, "time": 1792444885.5065198, "scalars": {"train/loss": 0.03567350655794144, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.7008354663848877e-08, "train/kl": 1.7836768627166748, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -2.9802322387695312e-08, "train/step_time_ms": 68.91226768493652, "train/samples_per_sec": 464.3585398510236, "train/data_wait_ms": 0.07104999997409323, "train/accepted": 1.0, "train/grad_norm": 0.8538321852684021, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.039682049000020925, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.4866071343421936, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 27, "time": 1792444885.5510402, "scalars": {"train/loss": 0.03480246663093567, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.7474015951156616e-08, "train/kl": 1.7401162385940552, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 1.4156103134155273e-07, "train/step_time_ms": 44.54994201660156, "train/samples_per_sec": 718.294986513679, "train/data_wait_ms": 0.08254100004023712, "train/accepted": 1.0, "train/grad_norm": 1.2680381536483765, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.041825098999993315, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.455357164144516, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 28, "time": 1792444885.6009095, "scalars": {"train/loss": 0.03761861100792885, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -9.12696123123169e-08, "train/kl": 1.880936622619629, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.1920928955078125e-07, "train/step_time_ms": 49.89004135131836, "train/samples_per_sec": 641.4105727966968, "train/data_wait_ms": 0.051649999988967465, "train/accepted": 1.0, "train/grad_norm": 1.142783522605896, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.037104467000006025, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.4955357313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 29, "time": 1792444885.6756496, "scalars": {"train/loss": 0.0407768189907074, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 4.6100467443466187e-08, "train/kl": 2.0388474464416504, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.30385160446167e-07, "train/step_time_ms": 74.7826099395752, "train/samples_per_sec": 427.9069696264438, "train/data_wait_ms": 0.09499000003643232, "train/accepted": 1.0, "train/grad_norm": 0.9434267282485962, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.043336950000025354, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.4955357313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 30, "time": 1792444885.7337642, "scalars": {"train/loss": 0.038389261811971664, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 4.563480615615845e-08, "train/kl": 1.9194623231887817, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 1.4901161193847656e-08, "train/step_time_ms": 58.090925216674805, "train/samples_per_sec": 550.8605669584775, "train/data_wait_ms": 0.1082100000076025, "train/accepted": 1.0, "train/grad_norm": 1.1257325410842896, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06979387299998052, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.4598214328289032, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 31, "time": 1792444885.7757254, "scalars": {"train/loss": 0.040212906897068024, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.6298145055770874e-08, "train/kl": 2.0106465816497803, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -2.2351741790771484e-08, "train/step_time_ms": 42.052507400512695, "train/samples_per_sec": 760.95343602769, "train/data_wait_ms": 0.04180000001952067, "train/accepted": 1.0, "train/grad_norm": 1.113492727279663, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.05211052499998914, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.4821428656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 32, "time": 1792444885.8413239, "scalars": {"train/loss": 0.04265898838639259, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 7.450580596923828e-09, "train/kl": 2.13295316696167, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -7.450580596923828e-08, "train/step_time_ms": 65.3986930847168, "train/samples_per_sec": 489.3064140983296, "train/data_wait_ms": 0.05542999997487641, "train/accepted": 1.0, "train/grad_norm": 0.7513960599899292, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.036974917999998524, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.5044642686843872, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 33, "time": 1792444885.882514, "scalars": {"train/loss": 0.04099344462156296, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -9.313225746154785e-10, "train/kl": 2.049670457839966, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 3.725290298461914e-08, "train/step_time_ms": 41.29147529602051, "train/samples_per_sec": 774.9783646767404, "train/data_wait_ms": 0.043259999984002206, "train/accepted": 1.0, "train/grad_norm": 0.7091848254203796, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.05996415800001387, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.5267857313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 34, "time": 1792444885.9207838, "scalars": {"train/loss": 0.039368029683828354, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.514570951461792e-08, "train/kl": 1.9683961868286133, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 1.0803341865539551e-07, "train/step_time_ms": 38.34700584411621, "train/samples_per_sec": 834.4849694414912, "train/data_wait_ms": 0.054910000017116545, "train/accepted": 1.0, "train/grad_norm": 0.45867636799812317, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03586679699998285, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.5625, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 35, "time": 1792444885.972587, "scalars": {"train/loss": 0.04016025736927986, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.0023435354232788e-08, "train/kl": 2.0080137252807617, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.4901161193847656e-08, "train/step_time_ms": 51.820993423461914, "train/samples_per_sec": 617.5103541243967, "train/data_wait_ms": 0.053530000002410816, "train/accepted": 1.0, "train/grad_norm": 0.7271668910980225, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03448105600000417, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.5178571939468384, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 36, "time": 1792444886.022572, "scalars": {"train/loss": 0.04589468613266945, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -9.313225746154785e-09, "train/kl": 2.294736385345459, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -4.0978193283081055e-08, "train/step_time_ms": 49.94320869445801, "train/samples_per_sec": 640.727755314426, "train/data_wait_ms": 0.052709999977196276, "train/accepted": 1.0, "train/grad_norm": 0.4863835573196411, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04799327300003142, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.5580357313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 37, "time": 1792444886.0649695, "scalars": {"train/loss": 0.03459109365940094, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 6.51925802230835e-09, "train/kl": 1.7295453548431396, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 1.862645149230957e-07, "train/step_time_ms": 42.20867156982422, "train/samples_per_sec": 758.138051017872, "train/data_wait_ms": 0.05228000003398847, "train/accepted": 1.0, "train/grad_norm": 0.6822102665901184, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04554237100001046, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.53125, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 38, "time": 1792444886.1352136, "scalars": {"train/loss": 0.041777074337005615, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 3.91155481338501e-08, "train/kl": 2.088860511779785, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.341104507446289e-07, "train/step_time_ms": 70.3744888305664, "train/samples_per_sec": 454.7102299676121, "train/data_wait_ms": 0.04949000003762194, "train/accepted": 1.0, "train/grad_norm": 0.6148926019668579, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03762477800000852, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.5491071939468384, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 39, "time": 1792444886.178189, "scalars": {"train/loss": 0.038581881672143936, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 5.587935447692871e-09, "train/kl": 1.9291030168533325, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.7881393432617188e-07, "train/step_time_ms": 42.97304153442383, "train/samples_per_sec": 744.6528999900135, "train/data_wait_ms": 0.06571000000121785, "train/accepted": 1.0, "train/grad_norm": 0.6614132523536682, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06483771099999558, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.5401785969734192, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 40, "time": 1792444886.2215753, "scalars": {"train/loss": 0.03566382825374603, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -3.725290298461914e-09, "train/kl": 1.7831909656524658, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 1.1175870895385742e-08, "train/step_time_ms": 43.471336364746094, "train/samples_per_sec": 736.1172366891166, "train/data_wait_ms": 0.07672999998931118, "train/accepted": 1.0, "train/grad_norm": 0.696683406829834, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.038268178999999236, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.6071428656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 41, "time": 1792444886.2837887, "scalars": {"train/loss": 0.037490569055080414, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.7939677238464355e-09, "train/kl": 1.8745315074920654, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -5.960464477539063e-08, "train/step_time_ms": 62.22414970397949, "train/samples_per_sec": 514.2697835524375, "train/data_wait_ms": 0.06744000000935557, "train/accepted": 1.0, "train/grad_norm": 0.5934330821037292, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.038377878000005694, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.5625, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 42, "time": 1792444886.3354325, "scalars": {"train/loss": 0.03716336563229561, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 4.190951585769653e-08, "train/kl": 1.8581764698028564, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.6391277313232422e-07, "train/step_time_ms": 51.52153968811035, "train/samples_per_sec": 621.099450709635, "train/data_wait_ms": 0.05894000003081601, "train/accepted": 1.0, "train/grad_norm": 0.577803373336792, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.05875322800000049, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.566964328289032, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 43, "time": 1792444886.378067, "scalars": {"train/loss": 0.032163895666599274, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 9.313225746154785e-09, "train/kl": 1.6081877946853638, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 1.4156103134155273e-07, "train/step_time_ms": 42.72961616516113, "train/samples_per_sec": 748.8950959987948, "train/data_wait_ms": 0.04526000003579611, "train/accepted": 1.0, "train/grad_norm": 0.6047043204307556, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04653162199997496, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.5892857313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 44, "time": 1792444886.4458919, "scalars": {"train/loss": 0.03271477669477463, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.60770320892334e-08, "train/kl": 1.6357359886169434, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 5.587935447692871e-08, "train/step_time_ms": 67.68584251403809, "train/samples_per_sec": 472.77242642526284, "train/data_wait_ms": 0.059789999966142204, "train/accepted": 1.0, "train/grad_norm": 0.8904069066047668, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.0383931979999943, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.5535714626312256, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 45, "time": 1792444886.492574, "scalars": {"train/loss": 0.03445553407073021, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -6.612390279769897e-08, "train/kl": 1.722773790359497, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 5.774199962615967e-08, "train/step_time_ms": 46.743154525756836, "train/samples_per_sec": 684.5922215704777, "train/data_wait_ms": 0.03982000004043584, "train/accepted": 1.0, "train/grad_norm": 0.8388816714286804, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06144166899997572, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.5982142686843872, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 46, "time": 1792444886.5346224, "scalars": {"train/loss": 0.03647478669881821, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.909211277961731e-08, "train/kl": 1.8237273693084717, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 2.384185791015625e-07, "train/step_time_ms": 42.14954376220703, "train/samples_per_sec": 759.2015747675182, "train/data_wait_ms": 0.0760300000024472, "train/accepted": 1.0, "train/grad_norm": 0.7195366621017456, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.040886699000054705, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.625, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 47, "time": 1792444886.607513, "scalars": {"train/loss": 0.03363696485757828, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -5.587935447692871e-09, "train/kl": 1.681849718093872, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -2.9802322387695312e-08, "train/step_time_ms": 72.61872291564941, "train/samples_per_sec": 440.6577080289574, "train/data_wait_ms": 0.04510000002255765, "train/accepted": 1.0, "train/grad_norm": 1.1805360317230225, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.037886367999988124, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.6383928656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 48, "time": 1792444886.6485803, "scalars": {"train/loss": 0.030635492876172066, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -3.166496753692627e-08, "train/kl": 1.531774640083313, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 41.35727882385254, "train/samples_per_sec": 773.7452973222264, "train/data_wait_ms": 0.0731100000166407, "train/accepted": 1.0, "train/grad_norm": 0.6144660115242004, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06644048099997235, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.6428571939468384, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 49, "time": 1792444886.6912549, "scalars": {"train/loss": 0.03280806913971901, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.3283064365386963e-08, "train/kl": 1.6404051780700684, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -3.3527612686157227e-08, "train/step_time_ms": 42.603492736816406, "train/samples_per_sec": 751.1121258925973, "train/data_wait_ms": 0.09476100001393206, "train/accepted": 1.0, "train/grad_norm": 0.6902381777763367, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03732898699996667, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.65625, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 50, "time": 1792444886.7714596, "scalars": {"train/loss": 0.03764590248465538, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.8870999813079834e-08, "train/kl": 1.8822989463806152, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -7.450580596923828e-08, "train/step_time_ms": 80.06739616394043, "train/samples_per_sec": 399.6633028315174, "train/data_wait_ms": 0.03789999999526117, "train/accepted": 1.0, "train/grad_norm": 0.8236232399940491, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03848121800001536, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.6607142686843872, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 51, "time": 1792444886.8119686, "scalars": {"train/loss": 0.038341961801052094, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.7939677238464355e-09, "train/kl": 1.917099952697754, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -3.725290298461914e-08, "train/step_time_ms": 40.66658020019531, "train/samples_per_sec": 786.8869190000469, "train/data_wait_ms": 0.06117100002711595, "train/accepted": 1.0, "train/grad_norm": 0.6816196441650391, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.07446413499997107, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.6696428656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 52, "time": 1792444886.8537843, "scalars": {"train/loss": 0.03582427278161049, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.6763806343078613e-08, "train/kl": 1.7912108898162842, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 5.4016709327697754e-08, "train/step_time_ms": 41.82887077331543, "train/samples_per_sec": 765.0218475516265, "train/data_wait_ms": 0.06560000002764355, "train/accepted": 1.0, "train/grad_norm": 0.8066308498382568, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.036325266999995165, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.6830357313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 53, "time": 1792444886.9189587, "scalars": {"train/loss": 0.03599347174167633, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.6763806343078613e-08, "train/kl": 1.7996752262115479, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -3.166496753692627e-08, "train/step_time_ms": 65.20223617553711, "train/samples_per_sec": 490.78071362230236, "train/data_wait_ms": 0.03691000000571876, "train/accepted": 1.0, "train/grad_norm": 0.6842591166496277, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03813830799998641, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.6741071939468384, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 54, "time": 1792444886.9562514, "scalars": {"train/loss": 0.039144210517406464, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.2351741790771484e-08, "train/kl": 1.9572150707244873, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -8.940696716308594e-08, "train/step_time_ms": 37.30130195617676, "train/samples_per_sec": 857.8789029293142, "train/data_wait_ms": 0.05234999997583145, "train/accepted": 1.0, "train/grad_norm": 0.5881405472755432, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06153743899994879, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7053571939468384, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 55, "time": 1792444886.9971797, "scalars": {"train/loss": 0.04115350916981697, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.1175870895385742e-08, "train/kl": 2.057682514190674, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.3969838619232178e-07, "train/step_time_ms": 40.92121124267578, "train/samples_per_sec": 781.9905381155469, "train/data_wait_ms": 0.08471000001009088, "train/accepted": 1.0, "train/grad_norm": 0.4407733380794525, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03362636599996449, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7232142686843872, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 56, "time": 1792444887.0555272, "scalars": {"train/loss": 0.036457013338804245, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.4901161193847656e-08, "train/kl": 1.8228516578674316, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.862645149230957e-08, "train/step_time_ms": 58.20584297180176, "train/samples_per_sec": 549.7729843978487, "train/data_wait_ms": 0.04288000002361514, "train/accepted": 1.0, "train/grad_norm": 0.3841986060142517, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03718380700001944, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.71875, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 57, "time": 1792444887.1022892, "scalars": {"train/loss": 0.03856976702809334, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -4.0046870708465576e-08, "train/kl": 1.9284858703613281, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 5.21540641784668e-08, "train/step_time_ms": 46.83709144592285, "train/samples_per_sec": 683.2191968398923, "train/data_wait_ms": 0.05234999997583145, "train/accepted": 1.0, "train/grad_norm": 0.49363648891448975, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.054239116000019294, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7142857313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 58, "time": 1792444887.1477609, "scalars": {"train/loss": 0.037086211144924164, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.955777406692505e-08, "train/kl": 1.8543155193328857, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -9.685754776000977e-08, "train/step_time_ms": 45.537710189819336, "train/samples_per_sec": 702.7142969334918, "train/data_wait_ms": 0.04615000000285363, "train/accepted": 1.0, "train/grad_norm": 0.3762725591659546, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04203660999996828, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7321428060531616, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 59, "time": 1792444887.2137997, "scalars": {"train/loss": 0.044077273458242416, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.6763806343078613e-08, "train/kl": 2.203866720199585, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -5.960464477539063e-08, "train/step_time_ms": 66.01166725158691, "train/samples_per_sec": 484.7627901601095, "train/data_wait_ms": 0.05103000000872271, "train/accepted": 1.0, "train/grad_norm": 0.5513920187950134, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04122066000002178, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7142857313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 60, "time": 1792444887.2577014, "scalars": {"train/loss": 0.041271403431892395, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -3.5390257835388184e-08, "train/kl": 2.0635743141174316, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -8.195638656616211e-08, "train/step_time_ms": 43.71166229248047, "train/samples_per_sec": 732.070077451729, "train/data_wait_ms": 0.04546999997501189, "train/accepted": 1.0, "train/grad_norm": 0.6451808214187622, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.061852429999987635, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.75, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 61, "time": 1792444887.3078763, "scalars": {"train/loss": 0.039687953889369965, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -9.313225746154785e-09, "train/kl": 1.984393835067749, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 7.82310962677002e-08, "train/step_time_ms": 50.25434494018555, "train/samples_per_sec": 636.7608619331821, "train/data_wait_ms": 0.05120999998098341, "train/accepted": 1.0, "train/grad_norm": 0.5234792828559875, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03913991899997882, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7544642686843872, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 62, "time": 1792444887.3691878, "scalars": {"train/loss": 0.038593415170907974, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.5832483768463135e-08, "train/kl": 1.9296741485595703, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -6.705522537231445e-08, "train/step_time_ms": 61.3250732421875, "train/samples_per_sec": 521.8094053247077, "train/data_wait_ms": 0.0786300000186202, "train/accepted": 1.0, "train/grad_norm": 0.41153037548065186, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04479517099997565, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.78125, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 63, "time": 1792444887.4117205, "scalars": {"train/loss": 0.036097753793001175, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.1420419216156006e-08, "train/kl": 1.8048911094665527, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -6.705522537231445e-08, "train/step_time_ms": 42.62042045593262, "train/samples_per_sec": 750.8138037513355, "train/data_wait_ms": 0.047080999991067074, "train/accepted": 1.0, "train/grad_norm": 0.6148787140846252, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.056059806000007484, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7633928656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 64, "time": 1792444887.4545012, "scalars": {"train/loss": 0.033066391944885254, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -4.470348358154297e-08, "train/kl": 1.6533210277557373, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -2.60770320892334e-08, "train/step_time_ms": 42.76275634765625, "train/samples_per_sec": 748.3147190008921, "train/data_wait_ms": 0.05175999996254177, "train/accepted": 1.0, "train/grad_norm": 0.21836915612220764, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03818561799999998, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7366071343421936, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 65, "time": 1792444887.5184991, "scalars": {"train/loss": 0.042623501271009445, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.8870999813079834e-08, "train/kl": 2.131182909011841, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.564621925354004e-07, "train/step_time_ms": 64.04328346252441, "train/samples_per_sec": 499.6620764880853, "train/data_wait_ms": 0.056710000023940665, "train/accepted": 1.0, "train/grad_norm": 0.6738010048866272, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.038693588000001, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7232142686843872, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 66, "time": 1792444887.5643418, "scalars": {"train/loss": 0.045055802911520004, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.0244548320770264e-08, "train/kl": 2.2527947425842285, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -8.940696716308594e-08, "train/step_time_ms": 45.815229415893555, "train/samples_per_sec": 698.4577051773755, "train/data_wait_ms": 0.05194999999957872, "train/accepted": 1.0, "train/grad_norm": 0.2717460095882416, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.05983367900000758, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7455357313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 67, "time": 1792444887.6066418, "scalars": {"train/loss": 0.03548663854598999, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.862645149230957e-09, "train/kl": 1.7743397951126099, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.564621925354004e-07, "train/step_time_ms": 42.17338562011719, "train/samples_per_sec": 758.7723757405817, "train/data_wait_ms": 0.08597999999437889, "train/accepted": 1.0, "train/grad_norm": 0.518161416053772, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.041534630000001016, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7589285373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 68, "time": 1792444887.6786609, "scalars": {"train/loss": 0.04558880254626274, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.0489096641540527e-08, "train/kl": 2.279438018798828, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 4.470348358154297e-08, "train/step_time_ms": 71.88630104064941, "train/samples_per_sec": 445.147399946271, "train/data_wait_ms": 0.05534000001716777, "train/accepted": 1.0, "train/grad_norm": 0.5199867486953735, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03797826800001758, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7589285969734192, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 69, "time": 1792444887.7205029, "scalars": {"train/loss": 0.042652953416109085, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 3.725290298461914e-09, "train/kl": 2.132643222808838, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 8.940696716308594e-08, "train/step_time_ms": 42.15288162231445, "train/samples_per_sec": 759.1414576758182, "train/data_wait_ms": 0.060039999993932724, "train/accepted": 1.0, "train/grad_norm": 0.31995177268981934, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06583403099995166, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8080357313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 70, "time": 1792444887.766605, "scalars": {"train/loss": 0.037905845791101456, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.30385160446167e-08, "train/kl": 1.8952935934066772, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -2.60770320892334e-08, "train/step_time_ms": 46.02789878845215, "train/samples_per_sec": 695.230519800057, "train/data_wait_ms": 0.05716999999094696, "train/accepted": 1.0, "train/grad_norm": 0.26969286799430847, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03770126799997797, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7633928656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 71, "time": 1792444887.8358212, "scalars": {"train/loss": 0.04260030388832092, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -3.725290298461914e-09, "train/kl": 2.13002872467041, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -2.682209014892578e-07, "train/step_time_ms": 69.00715827941895, "train/samples_per_sec": 463.7200081537606, "train/data_wait_ms": 0.040740000031291856, "train/accepted": 1.0, "train/grad_norm": 0.8281282782554626, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04227269999995542, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7767857313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 72, "time": 1792444887.8820622, "scalars": {"train/loss": 0.033942826092243195, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 3.4458935260772705e-08, "train/kl": 1.6971423625946045, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.862645149230957e-08, "train/step_time_ms": 46.24819755554199, "train/samples_per_sec": 691.9188571958821, "train/data_wait_ms": 0.045119999981579895, "train/accepted": 1.0, "train/grad_norm": 0.22486259043216705, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06335894000000053, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7857142686843872, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 73, "time": 1792444887.9528146, "scalars": {"train/loss": 0.03909550607204437, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.7939677238464355e-09, "train/kl": 1.954774260520935, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 2.2351741790771484e-08, "train/step_time_ms": 71.02060317993164, "train/samples_per_sec": 450.5734754030119, "train/data_wait_ms": 0.0828899999874011, "train/accepted": 1.0, "train/grad_norm": 0.3946702182292938, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04053104000001895, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7946428656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 74, "time": 1792444887.980552, "scalars": {"train/loss": 0.03646906092762947, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -5.587935447692871e-09, "train/kl": 1.8234546184539795, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -3.166496753692627e-08, "train/step_time_ms": 0.9288787841796875, "train/samples_per_sec": 34450.135523613964, "train/data_wait_ms": 0.0663299999814626, "train/accepted": 1.0, "train/grad_norm": 0.4308335483074188, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04628136200000199, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7946428656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 75, "time": 1792444888.0222242, "scalars": {"train/loss": 0.042169444262981415, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.4901161193847656e-08, "train/kl": 2.1084702014923096, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 4.0978193283081055e-08, "train/step_time_ms": 68.44639778137207, "train/samples_per_sec": 467.5191249978229, "train/data_wait_ms": 0.036399999999048305, "train/accepted": 1.0, "train/grad_norm": 0.2584823966026306, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.024259420999953818, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7946428656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 76, "time": 1792444888.063016, "scalars": {"train/loss": 0.036949966102838516, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.4901161193847656e-08, "train/kl": 1.8475021123886108, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -7.636845111846924e-08, "train/step_time_ms": 40.816545486450195, "train/samples_per_sec": 783.9957943188257, "train/data_wait_ms": 0.05679999998164931, "train/accepted": 1.0, "train/grad_norm": 0.1939679980278015, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.038215518000015436, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8348214626312256, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 77, "time": 1792444888.122951, "scalars": {"train/loss": 0.037769872695207596, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.6763806343078613e-08, "train/kl": 1.8884921073913574, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 2.9802322387695312e-08, "train/step_time_ms": 59.945106506347656, "train/samples_per_sec": 533.8217223220962, "train/data_wait_ms": 0.050579999992805824, "train/accepted": 1.0, "train/grad_norm": 0.5397806763648987, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03730378800003109, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7991071343421936, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 78, "time": 1792444888.1709738, "scalars": {"train/loss": 0.04063890129327774, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -3.5390257835388184e-08, "train/kl": 2.031949996948242, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -9.685754776000977e-08, "train/step_time_ms": 47.9731559753418, "train/samples_per_sec": 667.0397089665729, "train/data_wait_ms": 0.046359999998912826, "train/accepted": 1.0, "train/grad_norm": 0.4173940420150757, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.05696377799995389, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.78125, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 79, "time": 1792444888.2260747, "scalars": {"train/loss": 0.03560716286301613, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -3.166496753692627e-08, "train/kl": 1.7803583145141602, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.862645149230957e-09, "train/step_time_ms": 55.04250526428223, "train/samples_per_sec": 581.3688865787365, "train/data_wait_ms": 0.04533999998557192, "train/accepted": 1.0, "train/grad_norm": 0.36790159344673157, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.043054719999986446, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.75, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 80, "time": 1792444888.2933354, "scalars": {"train/loss": 0.0396047979593277, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.2351741790771484e-08, "train/kl": 1.9802433252334595, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -6.705522537231445e-08, "train/step_time_ms": 67.28601455688477, "train/samples_per_sec": 475.58174177408955, "train/data_wait_ms": 0.05624000004900154, "train/accepted": 1.0, "train/grad_norm": 0.25507980585098267, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.047641462999990836, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7946428656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 81, "time": 1792444888.336911, "scalars": {"train/loss": 0.034422170370817184, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -6.51925802230835e-09, "train/kl": 1.7211110591888428, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -4.842877388000488e-08, "train/step_time_ms": 43.587446212768555, "train/samples_per_sec": 734.1563404241354, "train/data_wait_ms": 0.04622000000154003, "train/accepted": 1.0, "train/grad_norm": 0.40108442306518555, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06180717000000868, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8125, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 82, "time": 1792444888.3792984, "scalars": {"train/loss": 0.034102894365787506, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 0.0, "train/kl": 1.7051498889923096, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.043081283569336e-07, "train/step_time_ms": 42.36602783203125, "train/samples_per_sec": 755.3221681973708, "train/data_wait_ms": 0.06415000001425142, "train/accepted": 1.0, "train/grad_norm": 0.4892493188381195, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.039007838999964406, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8214285969734192, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 83, "time": 1792444888.4487374, "scalars": {"train/loss": 0.04084628075361252, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.30385160446167e-08, "train/kl": 2.042320728302002, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.341104507446289e-07, "train/step_time_ms": 69.50521469116211, "train/samples_per_sec": 460.39711037780506, "train/data_wait_ms": 0.04678999999896405, "train/accepted": 1.0, "train/grad_norm": 0.7744091153144836, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.038024737999990066, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8080357313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 84, "time": 1792444888.4920008, "scalars": {"train/loss": 0.03978997841477394, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.0489096641540527e-08, "train/kl": 1.9894975423812866, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 2.7939677238464355e-08, "train/step_time_ms": 43.206214904785156, "train/samples_per_sec": 740.6341904867013, "train/data_wait_ms": 0.05128999998760264, "train/accepted": 1.0, "train/grad_norm": 0.09382984787225723, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06536926100000073, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7946428656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 85, "time": 1792444888.5330641, "scalars": {"train/loss": 0.033794134855270386, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.0489096641540527e-08, "train/kl": 1.689713954925537, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.4528632164001465e-07, "train/step_time_ms": 41.110992431640625, "train/samples_per_sec": 778.3806254059572, "train/data_wait_ms": 0.046929999996336846, "train/accepted": 1.0, "train/grad_norm": 0.25597161054611206, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.039211018999992575, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8125, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 86, "time": 1792444888.5960886, "scalars": {"train/loss": 0.04263667017221451, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.1175870895385742e-08, "train/kl": 2.131829023361206, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 8.940696716308594e-08, "train/step_time_ms": 63.0345344543457, "train/samples_per_sec": 507.65822698630035, "train/data_wait_ms": 0.09923999999728039, "train/accepted": 1.0, "train/grad_norm": 0.34462812542915344, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03706774800002677, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7901785373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 87, "time": 1792444888.6365306, "scalars": {"train/loss": 0.04717439413070679, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 0.0, "train/kl": 2.358719825744629, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 40.482282638549805, "train/samples_per_sec": 790.4692599899879, "train/data_wait_ms": 0.04318999998531581, "train/accepted": 1.0, "train/grad_norm": 0.35168755054473877, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.05895855799997207, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7455357313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 88, "time": 1792444888.6841042, "scalars": {"train/loss": 0.03659440577030182, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -3.725290298461914e-09, "train/kl": 1.8297231197357178, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -5.587935447692871e-08, "train/step_time_ms": 47.420501708984375, "train/samples_per_sec": 674.8136111334567, "train/data_wait_ms": 0.04208100000369086, "train/accepted": 1.0, "train/grad_norm": 0.9875986576080322, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03664286700001185, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.816964328289032, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 89, "time": 1792444888.755172, "scalars": {"train/loss": 0.03761148080229759, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.8870999813079834e-08, "train/kl": 1.8805735111236572, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 1.1175870895385742e-08, "train/step_time_ms": 71.11549377441406, "train/samples_per_sec": 449.97226766796297, "train/data_wait_ms": 0.053359999981239525, "train/accepted": 1.0, "train/grad_norm": 0.3347938060760498, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.0422129899999959, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7678571343421936, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 90, "time": 1792444888.796127, "scalars": {"train/loss": 0.04135102778673172, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 9.313225746154785e-09, "train/kl": 2.0675501823425293, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 2.421438694000244e-08, "train/step_time_ms": 41.04495048522949, "train/samples_per_sec": 779.6330516104673, "train/data_wait_ms": 0.06710999997494582, "train/accepted": 1.0, "train/grad_norm": 0.2733149230480194, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06708120200005396, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8348214030265808, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 91, "time": 1792444888.835879, "scalars": {"train/loss": 0.03542020544409752, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.9802322387695312e-08, "train/kl": 1.771010398864746, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 39.79659080505371, "train/samples_per_sec": 804.0889772883854, "train/data_wait_ms": 0.06039999999529755, "train/accepted": 1.0, "train/grad_norm": 0.0007835319847799838, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.0372913369999992, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8035714626312256, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 92, "time": 1792444888.8718245, "scalars": {"train/loss": 0.03820882365107536, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -7.450580596923828e-09, "train/kl": 1.910441279411316, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 35.87222099304199, "train/samples_per_sec": 892.055164529872, "train/data_wait_ms": 0.0579500000412736, "train/accepted": 1.0, "train/grad_norm": 0.0005605331971310079, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03680546699996512, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8035714626312256, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 93, "time": 1792444888.938127, "scalars": {"train/loss": 0.038806479424238205, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.1175870895385742e-08, "train/kl": 1.9403252601623535, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -2.421438694000244e-08, "train/step_time_ms": 27.05669403076172, "train/samples_per_sec": 1182.7017729371541, "train/data_wait_ms": 0.0403000000233078, "train/accepted": 1.0, "train/grad_norm": 0.12964768707752228, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03187059500004352, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8303571939468384, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 94, "time": 1792444888.9912834, "scalars": {"train/loss": 0.03691010922193527, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.421438694000244e-08, "train/kl": 1.8455016613006592, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 7.82310962677002e-08, "train/step_time_ms": 92.37265586853027, "train/samples_per_sec": 346.42286398633075, "train/data_wait_ms": 0.06715000000667715, "train/accepted": 1.0, "train/grad_norm": 0.1962258666753769, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.032717445000002954, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8258929252624512, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 95, "time": 1792444889.0554786, "scalars": {"train/loss": 0.03185807168483734, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.4901161193847656e-08, "train/kl": 1.5929036140441895, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 64.17155265808105, "train/samples_per_sec": 498.66332782225857, "train/data_wait_ms": 0.06992999999511085, "train/accepted": 1.0, "train/grad_norm": 0.0021711159497499466, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04816404299998567, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8392857313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 96, "time": 1792444889.098347, "scalars": {"train/loss": 0.03544877469539642, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 3.725290298461914e-08, "train/kl": 1.772437572479248, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 2.421438694000244e-08, "train/step_time_ms": 42.96588897705078, "train/samples_per_sec": 744.7768628060285, "train/data_wait_ms": 0.05095000000210348, "train/accepted": 1.0, "train/grad_norm": 0.09243104606866837, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.05946372800002564, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8526785373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 97, "time": 1792444889.1481023, "scalars": {"train/loss": 0.03825855627655983, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.862645149230957e-09, "train/kl": 1.9129278659820557, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 49.71599578857422, "train/samples_per_sec": 643.6560204101206, "train/data_wait_ms": 0.03787999997939551, "train/accepted": 1.0, "train/grad_norm": 0.0013153214240446687, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03899373799998784, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8214285373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 98, "time": 1792444889.212993, "scalars": {"train/loss": 0.03585898503661156, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.862645149230957e-09, "train/kl": 1.7929531335830688, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -7.82310962677002e-08, "train/step_time_ms": 64.91589546203613, "train/samples_per_sec": 492.94552239080053, "train/data_wait_ms": 0.04981000000725544, "train/accepted": 1.0, "train/grad_norm": 0.15233705937862396, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.046270511999978226, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7589285373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 99, "time": 1792444889.2524772, "scalars": {"train/loss": 0.04078877717256546, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 7.450580596923828e-09, "train/kl": 2.0394389629364014, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 39.408206939697266, "train/samples_per_sec": 812.0136003387985, "train/data_wait_ms": 0.041119999991678924, "train/accepted": 1.0, "train/grad_norm": 0.001566301449202001, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.061505968999995275, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8214285969734192, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 100, "time": 1792444889.3001838, "scalars": {"train/loss": 0.033499736338853836, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.6763806343078613e-08, "train/kl": 1.6749850511550903, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 3.5390257835388184e-08, "train/step_time_ms": 47.6384162902832, "train/samples_per_sec": 671.7267804414194, "train/data_wait_ms": 0.04546999997501189, "train/accepted": 1.0, "train/grad_norm": 0.7171792387962341, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03479789600004324, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8348214626312256, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 101, "time": 1792444889.366494, "scalars": {"train/loss": 0.03317442163825035, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 3.725290298461914e-09, "train/kl": 1.658721923828125, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.6763806343078613e-08, "train/step_time_ms": 66.39695167541504, "train/samples_per_sec": 481.94983643878214, "train/data_wait_ms": 0.04230000001825829, "train/accepted": 1.0, "train/grad_norm": 0.5336381196975708, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.042540859999974145, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8258928656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 102, "time": 1792444889.409794, "scalars": {"train/loss": 0.0435754656791687, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.4901161193847656e-08, "train/kl": 2.1787710189819336, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 4.470348358154297e-08, "train/step_time_ms": 43.27225685119629, "train/samples_per_sec": 739.503837528995, "train/data_wait_ms": 0.1928899999938949, "train/accepted": 1.0, "train/grad_norm": 0.33931419253349304, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.05884396800001923, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8348214626312256, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 103, "time": 1792444889.447461, "scalars": {"train/loss": 0.03488916903734207, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.2351741790771484e-08, "train/kl": 1.7444584369659424, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 37.74404525756836, "train/samples_per_sec": 847.8158549681006, "train/data_wait_ms": 0.08520000000089567, "train/accepted": 1.0, "train/grad_norm": 0.0033405539579689503, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03826120900004071, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7857142686843872, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 104, "time": 1792444889.5009592, "scalars": {"train/loss": 0.03375678136944771, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 3.5390257835388184e-08, "train/kl": 1.6878434419631958, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -8.754432201385498e-08, "train/step_time_ms": 53.429365158081055, "train/samples_per_sec": 598.9215837643185, "train/data_wait_ms": 0.07649000002629691, "train/accepted": 1.0, "train/grad_norm": 0.2149495631456375, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03411497599995528, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7991071939468384, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 105, "time": 1792444889.5444398, "scalars": {"train/loss": 0.03711199760437012, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 7.450580596923828e-09, "train/kl": 1.8555998802185059, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 43.50614547729492, "train/samples_per_sec": 735.5282719012703, "train/data_wait_ms": 0.04770999998982006, "train/accepted": 1.0, "train/grad_norm": 0.0018547078361734748, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.049398722999967504, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8392857313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 106, "time": 1792444889.586165, "scalars": {"train/loss": 0.03488829731941223, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 5.029141902923584e-08, "train/kl": 1.7444188594818115, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -7.82310962677002e-08, "train/step_time_ms": 41.724205017089844, "train/samples_per_sec": 766.9409156362141, "train/data_wait_ms": 0.04549999999881038, "train/accepted": 1.0, "train/grad_norm": 0.3637819290161133, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03919680900003186, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8526785373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 107, "time": 1792444889.6593382, "scalars": {"train/loss": 0.03690951690077782, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.4901161193847656e-08, "train/kl": 1.8454759120941162, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 73.08197021484375, "train/samples_per_sec": 437.8644952500261, "train/data_wait_ms": 0.04601000000548083, "train/accepted": 1.0, "train/grad_norm": 0.002759302733466029, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03743902799999432, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8214285373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 108, "time": 1792444889.7042503, "scalars": {"train/loss": 0.0443161204457283, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.7939677238464355e-08, "train/kl": 2.2158031463623047, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 5.960464477539063e-08, "train/step_time_ms": 44.89922523498535, "train/samples_per_sec": 712.7071755141487, "train/data_wait_ms": 0.05877999996073413, "train/accepted": 1.0, "train/grad_norm": 0.5490904450416565, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06735216200002014, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8080357313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 109, "time": 1792444889.7460177, "scalars": {"train/loss": 0.039440054446458817, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.1175870895385742e-08, "train/kl": 1.9720014333724976, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 2.7939677238464355e-08, "train/step_time_ms": 41.92161560058594, "train/samples_per_sec": 763.3293598434869, "train/data_wait_ms": 0.07811999995510632, "train/accepted": 1.0, "train/grad_norm": 0.16779953241348267, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.038869148000003406, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8348214626312256, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 110, "time": 1792444889.8092594, "scalars": {"train/loss": 0.040819015353918076, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.2351741790771484e-08, "train/kl": 2.0409507751464844, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 63.033342361450195, "train/samples_per_sec": 507.66782786962756, "train/data_wait_ms": 0.06951000000299246, "train/accepted": 1.0, "train/grad_norm": 0.0009136200533248484, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03806057799999962, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8035714626312256, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 111, "time": 1792444889.8586187, "scalars": {"train/loss": 0.034827154129743576, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.4901161193847656e-08, "train/kl": 1.7413578033447266, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 49.40199851989746, "train/samples_per_sec": 647.7470741818568, "train/data_wait_ms": 0.04303999998001018, "train/accepted": 1.0, "train/grad_norm": 0.0009702789830043912, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.057261906999997336, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8035714626312256, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 112, "time": 1792444889.9012716, "scalars": {"train/loss": 0.03622731938958168, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.4901161193847656e-08, "train/kl": 1.811366081237793, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 42.77944564819336, "train/samples_per_sec": 748.022783258095, "train/data_wait_ms": 0.10876000004600428, "train/accepted": 1.0, "train/grad_norm": 0.0007848817040212452, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04348394099997677, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8392857313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 113, "time": 1792444889.961522, "scalars": {"train/loss": 0.04375435784459114, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.7939677238464355e-08, "train/kl": 2.1877150535583496, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 5.774199962615967e-08, "train/step_time_ms": 60.212135314941406, "train/samples_per_sec": 531.4543294740009, "train/data_wait_ms": 0.07655999996813989, "train/accepted": 1.0, "train/grad_norm": 0.18343040347099304, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03811253799995029, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8482142686843872, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 114, "time": 1792444890.004667, "scalars": {"train/loss": 0.03851129487156868, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.862645149230957e-08, "train/kl": 1.9255695343017578, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -9.313225746154785e-08, "train/step_time_ms": 43.18952560424805, "train/samples_per_sec": 740.9203864200938, "train/data_wait_ms": 0.04450000000133514, "train/accepted": 1.0, "train/grad_norm": 0.12261220067739487, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.05613913700000239, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7991071939468384, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 115, "time": 1792444890.0522044, "scalars": {"train/loss": 0.03593607619404793, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.862645149230957e-09, "train/kl": 1.7968032360076904, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 1.1175870895385742e-08, "train/step_time_ms": 47.46866226196289, "train/samples_per_sec": 674.1289616168922, "train/data_wait_ms": 0.07032000002027416, "train/accepted": 1.0, "train/grad_norm": 0.15652835369110107, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03874828800002206, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8348214626312256, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 116, "time": 1792444890.123542, "scalars": {"train/loss": 0.03984593227505684, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.1175870895385742e-08, "train/kl": 1.9922974109649658, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.6763806343078613e-08, "train/step_time_ms": 71.33293151855469, "train/samples_per_sec": 448.60065777159815, "train/data_wait_ms": 0.05601999998816609, "train/accepted": 1.0, "train/grad_norm": 0.15158461034297943, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04174937999999884, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8258928656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 117, "time": 1792444890.1641905, "scalars": {"train/loss": 0.03991613909602165, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -7.450580596923828e-09, "train/kl": 1.9958069324493408, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 40.608882904052734, "train/samples_per_sec": 788.0049317191738, "train/data_wait_ms": 0.05941000000575514, "train/accepted": 1.0, "train/grad_norm": 0.0008253171108663082, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06665279199995666, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8214285373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 118, "time": 1792444890.212726, "scalars": {"train/loss": 0.03812960907816887, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.2351741790771484e-08, "train/kl": 1.9064805507659912, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 48.54536056518555, "train/samples_per_sec": 659.1773060791497, "train/data_wait_ms": 0.05373999999847001, "train/accepted": 1.0, "train/grad_norm": 0.6218665242195129, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03592989700001681, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8169642686843872, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 119, "time": 1792444890.2721224, "scalars": {"train/loss": 0.045419566333293915, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 6.984919309616089e-08, "train/kl": 2.2709765434265137, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 3.5390257835388184e-08, "train/step_time_ms": 59.558868408203125, "train/samples_per_sec": 537.2835457631461, "train/data_wait_ms": 0.0681199999803539, "train/accepted": 1.0, "train/grad_norm": 1.323392391204834, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04300720100002309, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8080357313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 120, "time": 1792444890.3066204, "scalars": {"train/loss": 0.03619974106550217, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.30385160446167e-08, "train/kl": 1.8099868297576904, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 5.587935447692871e-09, "train/step_time_ms": 34.49559211730957, "train/samples_per_sec": 927.6547534298649, "train/data_wait_ms": 0.057940000033340766, "train/accepted": 1.0, "train/grad_norm": 0.23149347305297852, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.05524085600001172, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8169642686843872, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 121, "time": 1792444890.3451262, "scalars": {"train/loss": 0.033356908708810806, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.9802322387695312e-08, "train/kl": 1.6678454875946045, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 38.39731216430664, "train/samples_per_sec": 833.3916671841043, "train/data_wait_ms": 0.044070000001283915, "train/accepted": 1.0, "train/grad_norm": 0.0005431281751953065, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03097811399999273, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8214285969734192, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 122, "time": 1792444890.382161, "scalars": {"train/loss": 0.03620756417512894, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.9802322387695312e-08, "train/kl": 1.8103781938552856, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 37.12129592895508, "train/samples_per_sec": 862.0388701203613, "train/data_wait_ms": 0.07544000004600093, "train/accepted": 1.0, "train/grad_norm": 0.0003540770267136395, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.0343989969999825, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8214285969734192, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 123, "time": 1792444890.4364903, "scalars": {"train/loss": 0.03676659241318703, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -3.725290298461914e-09, "train/kl": 1.838333010673523, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -6.705522537231445e-08, "train/step_time_ms": 54.26764488220215, "train/samples_per_sec": 589.6699602398787, "train/data_wait_ms": 0.0669599999696402, "train/accepted": 1.0, "train/grad_norm": 0.5360698103904724, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03342240599999968, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7946428656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 124, "time": 1792444890.4826772, "scalars": {"train/loss": 0.04522023722529411, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.60770320892334e-08, "train/kl": 2.2610116004943848, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 7.450580596923828e-09, "train/step_time_ms": 46.21386528015137, "train/samples_per_sec": 692.4328836381459, "train/data_wait_ms": 0.048389999960818386, "train/accepted": 1.0, "train/grad_norm": 0.2068033218383789, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.050494934000028024, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8035714626312256, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 125, "time": 1792444890.5207663, "scalars": {"train/loss": 0.03588353469967842, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -3.725290298461914e-08, "train/kl": 1.7941768169403076, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 38.14840316772461, "train/samples_per_sec": 838.8293438996037, "train/data_wait_ms": 0.0607300000297073, "train/accepted": 1.0, "train/grad_norm": 0.00027100867009721696, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04123795999998947, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8214285969734192, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 126, "time": 1792444890.5836055, "scalars": {"train/loss": 0.03398435562849045, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.9802322387695312e-08, "train/kl": 1.6992162466049194, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 3.166496753692627e-08, "train/step_time_ms": 62.70313262939453, "train/samples_per_sec": 510.3413283852226, "train/data_wait_ms": 0.05962000000181433, "train/accepted": 1.0, "train/grad_norm": 0.10506749898195267, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03463616699997374, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8348214626312256, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 127, "time": 1792444890.6242983, "scalars": {"train/loss": 0.035371728241443634, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.2351741790771484e-08, "train/kl": 1.7685863971710205, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 40.76695442199707, "train/samples_per_sec": 784.9494879787588, "train/data_wait_ms": 0.050350000037724385, "train/accepted": 1.0, "train/grad_norm": 0.0002748199622146785, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.05820528800001057, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8392857313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 128, "time": 1792444890.668154, "scalars": {"train/loss": 0.033463750034570694, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.862645149230957e-09, "train/kl": 1.673185110092163, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 5.029141902923584e-08, "train/step_time_ms": 43.72382164001465, "train/samples_per_sec": 731.8664929031414, "train/data_wait_ms": 0.09589000001142267, "train/accepted": 1.0, "train/grad_norm": 0.08234930038452148, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03642498799996474, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7767857313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 129, "time": 1792444890.733591, "scalars": {"train/loss": 0.03183818608522415, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.4901161193847656e-08, "train/kl": 1.591909408569336, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 65.58632850646973, "train/samples_per_sec": 487.9065611493008, "train/data_wait_ms": 0.0641099999825201, "train/accepted": 1.0, "train/grad_norm": 0.00310657755471766, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03831837799998539, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8214285373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 130, "time": 1792444890.7764363, "scalars": {"train/loss": 0.03602881357073784, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 7.450580596923828e-09, "train/kl": 1.8014397621154785, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 1.862645149230957e-08, "train/step_time_ms": 42.82259941101074, "train/samples_per_sec": 747.2689757308851, "train/data_wait_ms": 0.1384899999834488, "train/accepted": 1.0, "train/grad_norm": 1.2820460796356201, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.0612103590000288, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7991071939468384, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 131, "time": 1792444890.8175714, "scalars": {"train/loss": 0.03704780712723732, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -7.450580596923828e-09, "train/kl": 1.8523904085159302, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 41.11433029174805, "train/samples_per_sec": 778.317432703571, "train/data_wait_ms": 0.05787999998574378, "train/accepted": 1.0, "train/grad_norm": 0.0002554235397838056, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03879442800001698, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8392857313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 132, "time": 1792444890.898335, "scalars": {"train/loss": 0.035513050854206085, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.9802322387695312e-08, "train/kl": 1.775652527809143, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 80.78503608703613, "train/samples_per_sec": 396.11296287005257, "train/data_wait_ms": 0.05455000001575172, "train/accepted": 1.0, "train/grad_norm": 0.00042209221282973886, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03643467800003464, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8035714626312256, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 133, "time": 1792444890.9436007, "scalars": {"train/loss": 0.03458990529179573, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -5.587935447692871e-09, "train/kl": 1.729499340057373, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -8.009374141693115e-08, "train/step_time_ms": 45.258522033691406, "train/samples_per_sec": 707.0491602924753, "train/data_wait_ms": 0.054589999990639626, "train/accepted": 1.0, "train/grad_norm": 0.11006325483322144, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.0765779759999532, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7991071343421936, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 134, "time": 1792444890.98684, "scalars": {"train/loss": 0.03819416090846062, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 7.450580596923828e-09, "train/kl": 1.9097092151641846, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -2.421438694000244e-08, "train/step_time_ms": 43.26200485229492, "train/samples_per_sec": 739.6790811996428, "train/data_wait_ms": 0.053689999958805856, "train/accepted": 1.0, "train/grad_norm": 0.5634142160415649, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04169888000001265, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8258929252624512, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 135, "time": 1792444891.068998, "scalars": {"train/loss": 0.03893425688147545, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 4.470348358154297e-08, "train/kl": 1.9467129707336426, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 81.91156387329102, "train/samples_per_sec": 390.6652307298246, "train/data_wait_ms": 0.04546999997501189, "train/accepted": 1.0, "train/grad_norm": 0.0002589871874079108, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03941393899998502, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8035714626312256, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 136, "time": 1792444891.1138134, "scalars": {"train/loss": 0.041763171553611755, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 7.450580596923828e-09, "train/kl": 2.08815860748291, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 45.04227638244629, "train/samples_per_sec": 710.4436669295632, "train/data_wait_ms": 0.08041999996066806, "train/accepted": 1.0, "train/grad_norm": 0.00027166635845787823, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.0766453859999956, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8571428656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 137, "time": 1792444891.1613512, "scalars": {"train/loss": 0.0383259654045105, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 7.450580596923828e-09, "train/kl": 1.9163014888763428, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -6.51925802230835e-08, "train/step_time_ms": 47.486305236816406, "train/samples_per_sec": 673.8784969774867, "train/data_wait_ms": 0.09053999997377105, "train/accepted": 1.0, "train/grad_norm": 0.1211698055267334, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03984591900001533, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8526785373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 138, "time": 1792444891.22741, "scalars": {"train/loss": 0.0382532961666584, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -7.450580596923828e-09, "train/kl": 1.912663459777832, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 2.7939677238464355e-08, "train/step_time_ms": 66.0710334777832, "train/samples_per_sec": 484.3272205021615, "train/data_wait_ms": 0.04258000001300388, "train/accepted": 1.0, "train/grad_norm": 0.12884674966335297, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04176609000001008, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8125, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 139, "time": 1792444891.2746313, "scalars": {"train/loss": 0.039129581302404404, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.60770320892334e-08, "train/kl": 1.9564790725708008, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 47.1804141998291, "train/samples_per_sec": 678.2475428144061, "train/data_wait_ms": 0.0641099999825201, "train/accepted": 1.0, "train/grad_norm": 0.0003924043267033994, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06183037900001409, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8214285373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 140, "time": 1792444891.3579752, "scalars": {"train/loss": 0.0335196852684021, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 1.4901161193847656e-08, "train/kl": 1.6759843826293945, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 83.21285247802734, "train/samples_per_sec": 384.5559796000229, "train/data_wait_ms": 0.0432900000078007, "train/accepted": 1.0, "train/grad_norm": 0.00022701261332258582, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04212841000003209, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8392857313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 141, "time": 1792444891.4067688, "scalars": {"train/loss": 0.03846248984336853, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.2351741790771484e-08, "train/kl": 1.9231245517730713, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 1.8322467803955078, "train/samples_per_sec": 17464.896291476904, "train/data_wait_ms": 0.057450000042535976, "train/accepted": 1.0, "train/grad_norm": 0.0002658683806657791, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.0515029649999974, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8214285373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 142, "time": 1792444891.44984, "scalars": {"train/loss": 0.039944615215063095, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 3.725290298461914e-08, "train/kl": 1.9972307682037354, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 90.19088745117188, "train/samples_per_sec": 354.8030283804932, "train/data_wait_ms": 0.060539999992670346, "train/accepted": 1.0, "train/grad_norm": 0.0002780052600428462, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04308005000001458, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8214285373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 143, "time": 1792444891.517439, "scalars": {"train/loss": 0.032426703721284866, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -2.9802322387695312e-08, "train/kl": 1.6213352680206299, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 67.59834289550781, "train/samples_per_sec": 473.3843853164414, "train/data_wait_ms": 0.05631999999877735, "train/accepted": 1.0, "train/grad_norm": 0.00021323634427972138, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.037784117999990485, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8214285373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 144, "time": 1792444891.5627413, "scalars": {"train/loss": 0.03880034014582634, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -9.406358003616333e-08, "train/kl": 1.940018892288208, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -3.5390257835388184e-08, "train/step_time_ms": 45.352935791015625, "train/samples_per_sec": 705.5772562873244, "train/data_wait_ms": 0.06355000004987232, "train/accepted": 1.0, "train/grad_norm": 0.5582061409950256, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06246635000002243, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8080357313156128, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 145, "time": 1792444891.6014555, "scalars": {"train/loss": 0.04108047857880592, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -7.450580596923828e-09, "train/kl": 2.0540239810943604, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 38.78068923950195, "train/samples_per_sec": 825.1529466733883, "train/data_wait_ms": 0.06823000001077162, "train/accepted": 1.0, "train/grad_norm": 0.001315284869633615, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.04051667900000666, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8571428656578064, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 146, "time": 1792444891.6687634, "scalars": {"train/loss": 0.04112865775823593, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 7.450580596923828e-09, "train/kl": 2.0564329624176025, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 67.26741790771484, "train/samples_per_sec": 475.7132203870419, "train/data_wait_ms": 0.040630000000874134, "train/accepted": 1.0, "train/grad_norm": 0.0002864802081603557, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03371177600001829, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8035714626312256, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 147, "time": 1792444891.7112522, "scalars": {"train/loss": 0.03533254191279411, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 4.6566128730773926e-08, "train/kl": 1.7666270732879639, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 42.32478141784668, "train/samples_per_sec": 756.0582459737611, "train/data_wait_ms": 0.04671000004918824, "train/accepted": 1.0, "train/grad_norm": 0.0006109834066592157, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.06342433000003211, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.7857142686843872, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 148, "time": 1792444891.7498252, "scalars": {"train/loss": 0.045405466109514236, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -1.862645149230957e-09, "train/kl": 2.2702741622924805, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": -1.4901161193847656e-08, "train/step_time_ms": 38.72871398925781, "train/samples_per_sec": 826.2603299679881, "train/data_wait_ms": 0.047580000000380096, "train/accepted": 1.0, "train/grad_norm": 0.25397926568984985, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03675824700002295, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.84375, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 149, "time": 1792444891.8093076, "scalars": {"train/loss": 0.03397444263100624, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": -7.078051567077637e-08, "train/kl": 1.698722243309021, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 0.0, "train/step_time_ms": 59.087514877319336, "train/samples_per_sec": 541.5695695857257, "train/data_wait_ms": 0.07046000001764696, "train/accepted": 1.0, "train/grad_norm": 0.0036868194583803415, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.034305037000024186, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8214285373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 150, "time": 1792444891.824272, "scalars": {"train/loss": 0.04121430963277817, "train/actor/pg_clipfrac": 0.0, "train/actor/pg_clipfrac_lower": 0.0, "train/actor/ppo_kl": 2.0489096641540527e-08, "train/kl": 2.0607149600982666, "train/mismatch/ess": 1.0, "train/mismatch/k3_kl": 0.0, "train/mismatch/kl": 0.0, "train/pg": 1.30385160446167e-08, "train/step_time_ms": 15.187263488769531, "train/samples_per_sec": 2107.0286970172683, "train/data_wait_ms": 0.03979000001663735, "train/accepted": 1.0, "train/grad_norm": 0.3025514781475067, "train/compiles": 1.0, "train/cache_hits": 0.0, "train/cache_misses": 0.0, "train/learning_rate": 0.0010000000474974513, "train/rollout_seconds": 0.03457533700003523, "rollout/length/mean": 8.0, "rollout/reward/mean": 0.8526785373687744, "rollout/status/truncated": 1.0, "data/accepted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/attempted/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 512.0, "data/chunks_per_record/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0, "data/truncated/memory:ba5333df301c9ae60de68de5c585a95827443e7975a787385fc977ac66c8ba8c": 0.0}}
{"step": 150, "time": 1792444891.8604233, "scalars": {"goodput/time_to_first_step_s": 15.833273630000008, "goodput/step_fraction": 0.3281321087519973}}