RyanYr commited on
Commit
aad484c
1 Parent(s): 851c504

Training in progress, step 1200, checkpoint

Browse files
last-checkpoint/global_step1200/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:767f030ba2a5c63afc14cedb2e913f11942b3ad327633c0d74cb6fcffa46d603
3
+ size 24090788996
last-checkpoint/global_step1200/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6639149be195a31cd42b74ec21548b869ca6e8979806eb0691335ac775a8e99f
3
+ size 24090788996
last-checkpoint/global_step1200/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6984d80bac1f083018a62ddd2d32b09d30a4ccd663adbc284e4f8b34c00245a2
3
+ size 24090788996
last-checkpoint/global_step1200/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:55c61f846cc14b5e1459d751008e8c857405b3252496d056df51d783ed55f496
3
+ size 24090788996
last-checkpoint/global_step1200/zero_pp_rank_0_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e6792d5608674b0242de07a16aa4a221722d72b6d5e1a247c01485f39a41980b
3
+ size 150693
last-checkpoint/global_step1200/zero_pp_rank_1_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:92fe73515c49e01c09faa61d7ddac0f3a494172009b86a7048bb1d19469b114f
3
+ size 150693
last-checkpoint/global_step1200/zero_pp_rank_2_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:15dc71b409bb9383420625d20428c80c7a8be4f9022d7a1757f97135a1eeeb33
3
+ size 150693
last-checkpoint/global_step1200/zero_pp_rank_3_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:23da4697f71801d8d91e9ba8574e4e4c97ad107dc20889e0f11dfb01396cc6f0
3
+ size 150693
last-checkpoint/latest CHANGED
@@ -1 +1 @@
1
- global_step900
 
1
+ global_step1200
last-checkpoint/model-00001-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:14da53d37407b5d238ed5bed994d98ec29d72c40e049e4b1f2387678990f4e1b
3
  size 4976698672
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:490b89e78ff904b995825330f5ce8e5fa6c2b37c660e95d7b400021114917123
3
  size 4976698672
last-checkpoint/model-00002-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:006966f66153c92805b5b174f5befb28e60f6f9dcf3e1bd45fb96ab4d715fdb1
3
  size 4999802720
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fb7ffd46563ff9ab39f6aced4141ad3f324d906dff502b8a1abf7cad146a8847
3
  size 4999802720
last-checkpoint/model-00003-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4e8d56ccc537542e37a2ebab6bf4263ce54c3e1f27f1ac4f3e639447712cd51a
3
  size 4915916176
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f84c6354be1f2b2ad0cf9f00c6b4a71c832f6082e1980570655c9db0034d55c4
3
  size 4915916176
last-checkpoint/model-00004-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:124082e15566501c869453becb988cc88173d4683e7b94a703fa88832a7b7828
3
  size 1168138808
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:455aff8b0320fce7909efe3e2352281f10e0960882e59d328a0e7a06bfe38c09
3
  size 1168138808
last-checkpoint/rng_state_0.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bdd1f02cb20d3f4f7e0dd26fea62af57e5e71316163f926a28ed6cf89a9f3777
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8044e4c53158c210a17648ba8f2dc2d25a25bbfc55f686015542618eb652a33e
3
  size 15024
last-checkpoint/rng_state_1.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fc6d54ba2aa85e2f895439a1b787ec947b848a1c34ea5a3a28821572bf2b9fec
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4cd85d7fa425e7888c973f1c2985ac15ca21b5e6171fe140a401c2bc75ca46ff
3
  size 15024
last-checkpoint/rng_state_2.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8b6927d26551cddd8e35b34b43e79bd58f8b6027b6a481bb6a563a3652addeb4
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d7915667371a58f1598639e0d1c20a0c59c783c14580cd040a6631eb4ea2311e
3
  size 15024
last-checkpoint/rng_state_3.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f8295b1be8e66b4b30cb905dc48cfc717c027e427937b8142d00ae9de8106c6a
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:35dd78929ad7f0fbf37fdb1284e8edf0424350f6e6ce1cd5a3ee78979af3d3cb
3
  size 15024
last-checkpoint/scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e485be8a97b8384bfb5de699b32daa18a6f5e3237ab31d5e2e64251cc3fddac1
3
  size 1064
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fb6fabf84a4db93ed80ee4a419e3ff880be7088e879fe26d3b47309e87cf9f04
3
  size 1064
last-checkpoint/trainer_state.json CHANGED
@@ -1,9 +1,9 @@
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
- "epoch": 1.0262989095574087,
5
  "eval_steps": 100,
6
- "global_step": 900,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
@@ -6901,6 +6901,2304 @@
6901
  "eval_samples_per_second": 4.98,
6902
  "eval_steps_per_second": 1.245,
6903
  "step": 900
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6904
  }
6905
  ],
6906
  "logging_steps": 2,
 
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
+ "epoch": 1.3683985460765449,
5
  "eval_steps": 100,
6
+ "global_step": 1200,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
 
6901
  "eval_samples_per_second": 4.98,
6902
  "eval_steps_per_second": 1.245,
6903
  "step": 900
6904
+ },
6905
+ {
6906
+ "epoch": 1.0285795738008694,
6907
+ "grad_norm": 58.67281714259608,
6908
+ "learning_rate": 1.0456247275342074e-07,
6909
+ "logits/chosen": -1.3800104856491089,
6910
+ "logits/rejected": -1.4262235164642334,
6911
+ "logps/chosen": -148.66639709472656,
6912
+ "logps/rejected": -190.34461975097656,
6913
+ "loss": 0.4164,
6914
+ "rewards/accuracies": 0.75,
6915
+ "rewards/chosen": -0.14912372827529907,
6916
+ "rewards/margins": 0.7551414370536804,
6917
+ "rewards/rejected": -0.9042652249336243,
6918
+ "step": 902
6919
+ },
6920
+ {
6921
+ "epoch": 1.0308602380443304,
6922
+ "grad_norm": 51.15662895087173,
6923
+ "learning_rate": 1.0418249864123723e-07,
6924
+ "logits/chosen": -1.312926173210144,
6925
+ "logits/rejected": -1.3743692636489868,
6926
+ "logps/chosen": -210.05198669433594,
6927
+ "logps/rejected": -243.07321166992188,
6928
+ "loss": 0.446,
6929
+ "rewards/accuracies": 0.90625,
6930
+ "rewards/chosen": -0.018690135329961777,
6931
+ "rewards/margins": 0.9038247466087341,
6932
+ "rewards/rejected": -0.9225149154663086,
6933
+ "step": 904
6934
+ },
6935
+ {
6936
+ "epoch": 1.0331409022877913,
6937
+ "grad_norm": 60.88958208699774,
6938
+ "learning_rate": 1.0380246402632869e-07,
6939
+ "logits/chosen": -1.2510696649551392,
6940
+ "logits/rejected": -1.2662408351898193,
6941
+ "logps/chosen": -101.8481674194336,
6942
+ "logps/rejected": -151.2491912841797,
6943
+ "loss": 0.4442,
6944
+ "rewards/accuracies": 0.71875,
6945
+ "rewards/chosen": 0.025427617132663727,
6946
+ "rewards/margins": 0.740162193775177,
6947
+ "rewards/rejected": -0.7147345542907715,
6948
+ "step": 906
6949
+ },
6950
+ {
6951
+ "epoch": 1.0354215665312523,
6952
+ "grad_norm": 54.60415707659223,
6953
+ "learning_rate": 1.0342237440615764e-07,
6954
+ "logits/chosen": -1.3037830591201782,
6955
+ "logits/rejected": -1.3146332502365112,
6956
+ "logps/chosen": -202.0905303955078,
6957
+ "logps/rejected": -266.09552001953125,
6958
+ "loss": 0.414,
6959
+ "rewards/accuracies": 0.875,
6960
+ "rewards/chosen": -0.2531195878982544,
6961
+ "rewards/margins": 1.4251919984817505,
6962
+ "rewards/rejected": -1.6783115863800049,
6963
+ "step": 908
6964
+ },
6965
+ {
6966
+ "epoch": 1.0377022307747132,
6967
+ "grad_norm": 55.23259639285888,
6968
+ "learning_rate": 1.0304223527898243e-07,
6969
+ "logits/chosen": -1.3536947965621948,
6970
+ "logits/rejected": -1.371093988418579,
6971
+ "logps/chosen": -151.1886444091797,
6972
+ "logps/rejected": -162.5618438720703,
6973
+ "loss": 0.4135,
6974
+ "rewards/accuracies": 0.90625,
6975
+ "rewards/chosen": -0.06908969581127167,
6976
+ "rewards/margins": 0.756403386592865,
6977
+ "rewards/rejected": -0.8254930973052979,
6978
+ "step": 910
6979
+ },
6980
+ {
6981
+ "epoch": 1.039982895018174,
6982
+ "grad_norm": 55.36663265349372,
6983
+ "learning_rate": 1.0266205214377748e-07,
6984
+ "logits/chosen": -1.3304215669631958,
6985
+ "logits/rejected": -1.412177324295044,
6986
+ "logps/chosen": -165.882080078125,
6987
+ "logps/rejected": -179.88272094726562,
6988
+ "loss": 0.4386,
6989
+ "rewards/accuracies": 0.78125,
6990
+ "rewards/chosen": -0.23252317309379578,
6991
+ "rewards/margins": 0.8055883646011353,
6992
+ "rewards/rejected": -1.038111686706543,
6993
+ "step": 912
6994
+ },
6995
+ {
6996
+ "epoch": 1.042263559261635,
6997
+ "grad_norm": 57.72994301158152,
6998
+ "learning_rate": 1.0228183050015383e-07,
6999
+ "logits/chosen": -1.4341506958007812,
7000
+ "logits/rejected": -1.416825771331787,
7001
+ "logps/chosen": -149.43185424804688,
7002
+ "logps/rejected": -155.654296875,
7003
+ "loss": 0.4917,
7004
+ "rewards/accuracies": 0.78125,
7005
+ "rewards/chosen": -0.058750756084918976,
7006
+ "rewards/margins": 0.5951379537582397,
7007
+ "rewards/rejected": -0.6538887619972229,
7008
+ "step": 914
7009
+ },
7010
+ {
7011
+ "epoch": 1.044544223505096,
7012
+ "grad_norm": 55.93678999952222,
7013
+ "learning_rate": 1.0190157584827952e-07,
7014
+ "logits/chosen": -1.2692397832870483,
7015
+ "logits/rejected": -1.3736000061035156,
7016
+ "logps/chosen": -187.53558349609375,
7017
+ "logps/rejected": -219.8994903564453,
7018
+ "loss": 0.4593,
7019
+ "rewards/accuracies": 0.8125,
7020
+ "rewards/chosen": -0.10293256491422653,
7021
+ "rewards/margins": 0.9443311095237732,
7022
+ "rewards/rejected": -1.047263741493225,
7023
+ "step": 916
7024
+ },
7025
+ {
7026
+ "epoch": 1.0468248877485569,
7027
+ "grad_norm": 54.86148488530044,
7028
+ "learning_rate": 1.0152129368880021e-07,
7029
+ "logits/chosen": -1.359283208847046,
7030
+ "logits/rejected": -1.423500657081604,
7031
+ "logps/chosen": -160.22647094726562,
7032
+ "logps/rejected": -174.9389190673828,
7033
+ "loss": 0.4786,
7034
+ "rewards/accuracies": 0.875,
7035
+ "rewards/chosen": -0.20612327754497528,
7036
+ "rewards/margins": 0.813952624797821,
7037
+ "rewards/rejected": -1.0200759172439575,
7038
+ "step": 918
7039
+ },
7040
+ {
7041
+ "epoch": 1.0491055519920176,
7042
+ "grad_norm": 53.53882999084853,
7043
+ "learning_rate": 1.0114098952275933e-07,
7044
+ "logits/chosen": -1.4099137783050537,
7045
+ "logits/rejected": -1.4628063440322876,
7046
+ "logps/chosen": -158.679443359375,
7047
+ "logps/rejected": -185.23330688476562,
7048
+ "loss": 0.4698,
7049
+ "rewards/accuracies": 0.90625,
7050
+ "rewards/chosen": -0.0398465096950531,
7051
+ "rewards/margins": 0.869282603263855,
7052
+ "rewards/rejected": -0.9091290831565857,
7053
+ "step": 920
7054
+ },
7055
+ {
7056
+ "epoch": 1.0513862162354786,
7057
+ "grad_norm": 63.92382605562123,
7058
+ "learning_rate": 1.0076066885151877e-07,
7059
+ "logits/chosen": -1.2206496000289917,
7060
+ "logits/rejected": -1.3061589002609253,
7061
+ "logps/chosen": -192.36740112304688,
7062
+ "logps/rejected": -219.992431640625,
7063
+ "loss": 0.4498,
7064
+ "rewards/accuracies": 0.9375,
7065
+ "rewards/chosen": -0.20084695518016815,
7066
+ "rewards/margins": 1.0578863620758057,
7067
+ "rewards/rejected": -1.2587332725524902,
7068
+ "step": 922
7069
+ },
7070
+ {
7071
+ "epoch": 1.0536668804789395,
7072
+ "grad_norm": 58.361553235329055,
7073
+ "learning_rate": 1.0038033717667906e-07,
7074
+ "logits/chosen": -1.3777412176132202,
7075
+ "logits/rejected": -1.4997334480285645,
7076
+ "logps/chosen": -144.8574676513672,
7077
+ "logps/rejected": -185.2276611328125,
7078
+ "loss": 0.4444,
7079
+ "rewards/accuracies": 0.875,
7080
+ "rewards/chosen": -0.12493613362312317,
7081
+ "rewards/margins": 0.8183287978172302,
7082
+ "rewards/rejected": -0.943264901638031,
7083
+ "step": 924
7084
+ },
7085
+ {
7086
+ "epoch": 1.0559475447224005,
7087
+ "grad_norm": 60.9980807817603,
7088
+ "learning_rate": 1e-07,
7089
+ "logits/chosen": -1.3119468688964844,
7090
+ "logits/rejected": -1.4459898471832275,
7091
+ "logps/chosen": -204.2576904296875,
7092
+ "logps/rejected": -243.5587921142578,
7093
+ "loss": 0.4801,
7094
+ "rewards/accuracies": 0.78125,
7095
+ "rewards/chosen": -0.2800896167755127,
7096
+ "rewards/margins": 0.8604587912559509,
7097
+ "rewards/rejected": -1.1405483484268188,
7098
+ "step": 926
7099
+ },
7100
+ {
7101
+ "epoch": 1.0582282089658612,
7102
+ "grad_norm": 58.9377536477793,
7103
+ "learning_rate": 9.961966282332094e-08,
7104
+ "logits/chosen": -1.4508683681488037,
7105
+ "logits/rejected": -1.4245975017547607,
7106
+ "logps/chosen": -168.61058044433594,
7107
+ "logps/rejected": -165.25489807128906,
7108
+ "loss": 0.4758,
7109
+ "rewards/accuracies": 0.6875,
7110
+ "rewards/chosen": -0.2604736089706421,
7111
+ "rewards/margins": 0.6006544828414917,
7112
+ "rewards/rejected": -0.8611280918121338,
7113
+ "step": 928
7114
+ },
7115
+ {
7116
+ "epoch": 1.0605088732093222,
7117
+ "grad_norm": 56.86318861636952,
7118
+ "learning_rate": 9.923933114848123e-08,
7119
+ "logits/chosen": -1.1611130237579346,
7120
+ "logits/rejected": -1.1586978435516357,
7121
+ "logps/chosen": -137.10975646972656,
7122
+ "logps/rejected": -159.64285278320312,
7123
+ "loss": 0.4456,
7124
+ "rewards/accuracies": 0.78125,
7125
+ "rewards/chosen": -0.2737983167171478,
7126
+ "rewards/margins": 0.6588336825370789,
7127
+ "rewards/rejected": -0.9326320290565491,
7128
+ "step": 930
7129
+ },
7130
+ {
7131
+ "epoch": 1.0627895374527832,
7132
+ "grad_norm": 47.87292073584732,
7133
+ "learning_rate": 9.885901047724065e-08,
7134
+ "logits/chosen": -1.3931517601013184,
7135
+ "logits/rejected": -1.3711789846420288,
7136
+ "logps/chosen": -123.40711212158203,
7137
+ "logps/rejected": -148.42681884765625,
7138
+ "loss": 0.457,
7139
+ "rewards/accuracies": 0.9375,
7140
+ "rewards/chosen": 0.12486562132835388,
7141
+ "rewards/margins": 0.9808838963508606,
7142
+ "rewards/rejected": -0.8560183048248291,
7143
+ "step": 932
7144
+ },
7145
+ {
7146
+ "epoch": 1.0650702016962441,
7147
+ "grad_norm": 60.27486161031482,
7148
+ "learning_rate": 9.847870631119978e-08,
7149
+ "logits/chosen": -1.1859281063079834,
7150
+ "logits/rejected": -1.167513370513916,
7151
+ "logps/chosen": -141.37530517578125,
7152
+ "logps/rejected": -153.80101013183594,
7153
+ "loss": 0.4799,
7154
+ "rewards/accuracies": 0.8125,
7155
+ "rewards/chosen": -0.2075161635875702,
7156
+ "rewards/margins": 0.6786595582962036,
7157
+ "rewards/rejected": -0.886175811290741,
7158
+ "step": 934
7159
+ },
7160
+ {
7161
+ "epoch": 1.0673508659397049,
7162
+ "grad_norm": 52.567168828966985,
7163
+ "learning_rate": 9.809842415172047e-08,
7164
+ "logits/chosen": -1.322398066520691,
7165
+ "logits/rejected": -1.4248247146606445,
7166
+ "logps/chosen": -136.26565551757812,
7167
+ "logps/rejected": -173.91111755371094,
7168
+ "loss": 0.4799,
7169
+ "rewards/accuracies": 0.75,
7170
+ "rewards/chosen": -0.21634379029273987,
7171
+ "rewards/margins": 0.6690733432769775,
7172
+ "rewards/rejected": -0.8854170441627502,
7173
+ "step": 936
7174
+ },
7175
+ {
7176
+ "epoch": 1.0696315301831658,
7177
+ "grad_norm": 55.376994474776346,
7178
+ "learning_rate": 9.771816949984619e-08,
7179
+ "logits/chosen": -1.1317143440246582,
7180
+ "logits/rejected": -1.2357646226882935,
7181
+ "logps/chosen": -102.08535766601562,
7182
+ "logps/rejected": -140.366943359375,
7183
+ "loss": 0.4312,
7184
+ "rewards/accuracies": 0.875,
7185
+ "rewards/chosen": -0.09264838695526123,
7186
+ "rewards/margins": 0.8339763879776001,
7187
+ "rewards/rejected": -0.9266247153282166,
7188
+ "step": 938
7189
+ },
7190
+ {
7191
+ "epoch": 1.0719121944266268,
7192
+ "grad_norm": 52.064161742890775,
7193
+ "learning_rate": 9.733794785622253e-08,
7194
+ "logits/chosen": -1.2546545267105103,
7195
+ "logits/rejected": -1.2125529050827026,
7196
+ "logps/chosen": -137.27352905273438,
7197
+ "logps/rejected": -155.36073303222656,
7198
+ "loss": 0.429,
7199
+ "rewards/accuracies": 0.8125,
7200
+ "rewards/chosen": -0.31309816241264343,
7201
+ "rewards/margins": 0.9382134675979614,
7202
+ "rewards/rejected": -1.2513116598129272,
7203
+ "step": 940
7204
+ },
7205
+ {
7206
+ "epoch": 1.0741928586700877,
7207
+ "grad_norm": 54.0017692542014,
7208
+ "learning_rate": 9.695776472101757e-08,
7209
+ "logits/chosen": -1.444411039352417,
7210
+ "logits/rejected": -1.4067819118499756,
7211
+ "logps/chosen": -167.93966674804688,
7212
+ "logps/rejected": -159.4579315185547,
7213
+ "loss": 0.4581,
7214
+ "rewards/accuracies": 0.59375,
7215
+ "rewards/chosen": -0.14569707214832306,
7216
+ "rewards/margins": 0.46156030893325806,
7217
+ "rewards/rejected": -0.6072573661804199,
7218
+ "step": 942
7219
+ },
7220
+ {
7221
+ "epoch": 1.0764735229135485,
7222
+ "grad_norm": 59.11201512010301,
7223
+ "learning_rate": 9.657762559384235e-08,
7224
+ "logits/chosen": -1.2698876857757568,
7225
+ "logits/rejected": -1.3079754114151,
7226
+ "logps/chosen": -191.69493103027344,
7227
+ "logps/rejected": -204.89511108398438,
7228
+ "loss": 0.45,
7229
+ "rewards/accuracies": 0.78125,
7230
+ "rewards/chosen": -0.35640037059783936,
7231
+ "rewards/margins": 0.654572606086731,
7232
+ "rewards/rejected": -1.0109730958938599,
7233
+ "step": 944
7234
+ },
7235
+ {
7236
+ "epoch": 1.0787541871570094,
7237
+ "grad_norm": 62.61929808626616,
7238
+ "learning_rate": 9.619753597367133e-08,
7239
+ "logits/chosen": -1.3809797763824463,
7240
+ "logits/rejected": -1.465749979019165,
7241
+ "logps/chosen": -113.58399963378906,
7242
+ "logps/rejected": -143.52798461914062,
7243
+ "loss": 0.5093,
7244
+ "rewards/accuracies": 0.59375,
7245
+ "rewards/chosen": -0.23512893915176392,
7246
+ "rewards/margins": 0.46248409152030945,
7247
+ "rewards/rejected": -0.6976129412651062,
7248
+ "step": 946
7249
+ },
7250
+ {
7251
+ "epoch": 1.0810348514004704,
7252
+ "grad_norm": 62.384552822468294,
7253
+ "learning_rate": 9.581750135876276e-08,
7254
+ "logits/chosen": -1.156672477722168,
7255
+ "logits/rejected": -1.1879075765609741,
7256
+ "logps/chosen": -136.56668090820312,
7257
+ "logps/rejected": -164.78175354003906,
7258
+ "loss": 0.4693,
7259
+ "rewards/accuracies": 0.84375,
7260
+ "rewards/chosen": -0.1409253031015396,
7261
+ "rewards/margins": 0.768065333366394,
7262
+ "rewards/rejected": -0.9089905619621277,
7263
+ "step": 948
7264
+ },
7265
+ {
7266
+ "epoch": 1.0833155156439314,
7267
+ "grad_norm": 62.82871732010386,
7268
+ "learning_rate": 9.543752724657923e-08,
7269
+ "logits/chosen": -1.2612335681915283,
7270
+ "logits/rejected": -1.3249022960662842,
7271
+ "logps/chosen": -148.45864868164062,
7272
+ "logps/rejected": -204.80255126953125,
7273
+ "loss": 0.4975,
7274
+ "rewards/accuracies": 0.78125,
7275
+ "rewards/chosen": -0.337429940700531,
7276
+ "rewards/margins": 1.0719372034072876,
7277
+ "rewards/rejected": -1.4093672037124634,
7278
+ "step": 950
7279
+ },
7280
+ {
7281
+ "epoch": 1.0855961798873923,
7282
+ "grad_norm": 51.27361713332628,
7283
+ "learning_rate": 9.505761913370813e-08,
7284
+ "logits/chosen": -1.2637767791748047,
7285
+ "logits/rejected": -1.2903871536254883,
7286
+ "logps/chosen": -123.89924621582031,
7287
+ "logps/rejected": -151.03236389160156,
7288
+ "loss": 0.4743,
7289
+ "rewards/accuracies": 0.8125,
7290
+ "rewards/chosen": -0.15131911635398865,
7291
+ "rewards/margins": 0.7663697600364685,
7292
+ "rewards/rejected": -0.9176889061927795,
7293
+ "step": 952
7294
+ },
7295
+ {
7296
+ "epoch": 1.087876844130853,
7297
+ "grad_norm": 60.844362449035934,
7298
+ "learning_rate": 9.467778251578216e-08,
7299
+ "logits/chosen": -1.264873743057251,
7300
+ "logits/rejected": -1.308799147605896,
7301
+ "logps/chosen": -211.9695281982422,
7302
+ "logps/rejected": -249.72434997558594,
7303
+ "loss": 0.4411,
7304
+ "rewards/accuracies": 0.84375,
7305
+ "rewards/chosen": -0.4224779009819031,
7306
+ "rewards/margins": 1.1005442142486572,
7307
+ "rewards/rejected": -1.523022174835205,
7308
+ "step": 954
7309
+ },
7310
+ {
7311
+ "epoch": 1.090157508374314,
7312
+ "grad_norm": 62.233166596514444,
7313
+ "learning_rate": 9.429802288739961e-08,
7314
+ "logits/chosen": -1.2648420333862305,
7315
+ "logits/rejected": -1.3022546768188477,
7316
+ "logps/chosen": -204.7000732421875,
7317
+ "logps/rejected": -234.1602783203125,
7318
+ "loss": 0.5,
7319
+ "rewards/accuracies": 0.75,
7320
+ "rewards/chosen": -0.3554447889328003,
7321
+ "rewards/margins": 0.6820401549339294,
7322
+ "rewards/rejected": -1.037484884262085,
7323
+ "step": 956
7324
+ },
7325
+ {
7326
+ "epoch": 1.092438172617775,
7327
+ "grad_norm": 52.22595835402743,
7328
+ "learning_rate": 9.391834574204533e-08,
7329
+ "logits/chosen": -1.3593980073928833,
7330
+ "logits/rejected": -1.3903865814208984,
7331
+ "logps/chosen": -180.24501037597656,
7332
+ "logps/rejected": -211.7206268310547,
7333
+ "loss": 0.4345,
7334
+ "rewards/accuracies": 0.9375,
7335
+ "rewards/chosen": -0.10301375389099121,
7336
+ "rewards/margins": 1.223108172416687,
7337
+ "rewards/rejected": -1.3261220455169678,
7338
+ "step": 958
7339
+ },
7340
+ {
7341
+ "epoch": 1.0947188368612357,
7342
+ "grad_norm": 59.31222116238765,
7343
+ "learning_rate": 9.353875657201083e-08,
7344
+ "logits/chosen": -1.2963013648986816,
7345
+ "logits/rejected": -1.3758771419525146,
7346
+ "logps/chosen": -183.5509033203125,
7347
+ "logps/rejected": -246.8356475830078,
7348
+ "loss": 0.4517,
7349
+ "rewards/accuracies": 0.90625,
7350
+ "rewards/chosen": -0.12891785800457,
7351
+ "rewards/margins": 1.3549247980117798,
7352
+ "rewards/rejected": -1.4838427305221558,
7353
+ "step": 960
7354
+ },
7355
+ {
7356
+ "epoch": 1.0969995011046967,
7357
+ "grad_norm": 70.97226420908964,
7358
+ "learning_rate": 9.3159260868315e-08,
7359
+ "logits/chosen": -1.3708178997039795,
7360
+ "logits/rejected": -1.43119478225708,
7361
+ "logps/chosen": -188.970703125,
7362
+ "logps/rejected": -210.37744140625,
7363
+ "loss": 0.4582,
7364
+ "rewards/accuracies": 0.8125,
7365
+ "rewards/chosen": -0.3142687976360321,
7366
+ "rewards/margins": 0.9785380959510803,
7367
+ "rewards/rejected": -1.29280686378479,
7368
+ "step": 962
7369
+ },
7370
+ {
7371
+ "epoch": 1.0992801653481576,
7372
+ "grad_norm": 62.07148030594709,
7373
+ "learning_rate": 9.277986412062474e-08,
7374
+ "logits/chosen": -1.2510358095169067,
7375
+ "logits/rejected": -1.3639835119247437,
7376
+ "logps/chosen": -151.8282012939453,
7377
+ "logps/rejected": -186.52447509765625,
7378
+ "loss": 0.3958,
7379
+ "rewards/accuracies": 0.8125,
7380
+ "rewards/chosen": -0.29675284028053284,
7381
+ "rewards/margins": 0.9109016060829163,
7382
+ "rewards/rejected": -1.2076544761657715,
7383
+ "step": 964
7384
+ },
7385
+ {
7386
+ "epoch": 1.1015608295916186,
7387
+ "grad_norm": 68.63281362186014,
7388
+ "learning_rate": 9.240057181717546e-08,
7389
+ "logits/chosen": -1.199402093887329,
7390
+ "logits/rejected": -1.3185701370239258,
7391
+ "logps/chosen": -170.2873077392578,
7392
+ "logps/rejected": -201.01829528808594,
7393
+ "loss": 0.4737,
7394
+ "rewards/accuracies": 0.78125,
7395
+ "rewards/chosen": -0.3943040668964386,
7396
+ "rewards/margins": 0.6716153025627136,
7397
+ "rewards/rejected": -1.0659193992614746,
7398
+ "step": 966
7399
+ },
7400
+ {
7401
+ "epoch": 1.1038414938350796,
7402
+ "grad_norm": 53.596924936810275,
7403
+ "learning_rate": 9.202138944469167e-08,
7404
+ "logits/chosen": -1.3583130836486816,
7405
+ "logits/rejected": -1.3962550163269043,
7406
+ "logps/chosen": -138.78871154785156,
7407
+ "logps/rejected": -169.90960693359375,
7408
+ "loss": 0.4341,
7409
+ "rewards/accuracies": 0.78125,
7410
+ "rewards/chosen": -0.13780230283737183,
7411
+ "rewards/margins": 0.8456271886825562,
7412
+ "rewards/rejected": -0.9834293723106384,
7413
+ "step": 968
7414
+ },
7415
+ {
7416
+ "epoch": 1.1061221580785403,
7417
+ "grad_norm": 57.028026126969664,
7418
+ "learning_rate": 9.164232248830777e-08,
7419
+ "logits/chosen": -1.271955966949463,
7420
+ "logits/rejected": -1.3388372659683228,
7421
+ "logps/chosen": -176.40234375,
7422
+ "logps/rejected": -217.4737091064453,
7423
+ "loss": 0.4131,
7424
+ "rewards/accuracies": 0.8125,
7425
+ "rewards/chosen": -0.16253671050071716,
7426
+ "rewards/margins": 1.022351861000061,
7427
+ "rewards/rejected": -1.1848886013031006,
7428
+ "step": 970
7429
+ },
7430
+ {
7431
+ "epoch": 1.1084028223220013,
7432
+ "grad_norm": 69.9853718884853,
7433
+ "learning_rate": 9.126337643148838e-08,
7434
+ "logits/chosen": -1.2280056476593018,
7435
+ "logits/rejected": -1.2864279747009277,
7436
+ "logps/chosen": -188.2535858154297,
7437
+ "logps/rejected": -217.4619598388672,
7438
+ "loss": 0.4778,
7439
+ "rewards/accuracies": 0.8125,
7440
+ "rewards/chosen": -0.3671016991138458,
7441
+ "rewards/margins": 0.9746519923210144,
7442
+ "rewards/rejected": -1.3417538404464722,
7443
+ "step": 972
7444
+ },
7445
+ {
7446
+ "epoch": 1.1106834865654622,
7447
+ "grad_norm": 65.03116909989372,
7448
+ "learning_rate": 9.088455675594942e-08,
7449
+ "logits/chosen": -1.2964333295822144,
7450
+ "logits/rejected": -1.3419139385223389,
7451
+ "logps/chosen": -208.0177764892578,
7452
+ "logps/rejected": -250.48910522460938,
7453
+ "loss": 0.4301,
7454
+ "rewards/accuracies": 0.875,
7455
+ "rewards/chosen": -0.375073105096817,
7456
+ "rewards/margins": 1.1881136894226074,
7457
+ "rewards/rejected": -1.5631868839263916,
7458
+ "step": 974
7459
+ },
7460
+ {
7461
+ "epoch": 1.1129641508089232,
7462
+ "grad_norm": 53.40636924718595,
7463
+ "learning_rate": 9.050586894157853e-08,
7464
+ "logits/chosen": -1.2230535745620728,
7465
+ "logits/rejected": -1.3127639293670654,
7466
+ "logps/chosen": -152.9366455078125,
7467
+ "logps/rejected": -209.3880157470703,
7468
+ "loss": 0.4103,
7469
+ "rewards/accuracies": 0.75,
7470
+ "rewards/chosen": -0.21992121636867523,
7471
+ "rewards/margins": 1.1222590208053589,
7472
+ "rewards/rejected": -1.3421802520751953,
7473
+ "step": 976
7474
+ },
7475
+ {
7476
+ "epoch": 1.115244815052384,
7477
+ "grad_norm": 65.88010848155325,
7478
+ "learning_rate": 9.012731846635589e-08,
7479
+ "logits/chosen": -1.1757484674453735,
7480
+ "logits/rejected": -1.1423841714859009,
7481
+ "logps/chosen": -194.4547119140625,
7482
+ "logps/rejected": -219.962890625,
7483
+ "loss": 0.4572,
7484
+ "rewards/accuracies": 0.78125,
7485
+ "rewards/chosen": -0.47297030687332153,
7486
+ "rewards/margins": 0.9668577909469604,
7487
+ "rewards/rejected": -1.4398282766342163,
7488
+ "step": 978
7489
+ },
7490
+ {
7491
+ "epoch": 1.117525479295845,
7492
+ "grad_norm": 63.30481794169962,
7493
+ "learning_rate": 8.974891080627503e-08,
7494
+ "logits/chosen": -1.297542929649353,
7495
+ "logits/rejected": -1.387624740600586,
7496
+ "logps/chosen": -152.45774841308594,
7497
+ "logps/rejected": -171.50283813476562,
7498
+ "loss": 0.4435,
7499
+ "rewards/accuracies": 0.8125,
7500
+ "rewards/chosen": -0.17080621421337128,
7501
+ "rewards/margins": 0.6862813830375671,
7502
+ "rewards/rejected": -0.8570876717567444,
7503
+ "step": 980
7504
+ },
7505
+ {
7506
+ "epoch": 1.1198061435393059,
7507
+ "grad_norm": 58.37292774536043,
7508
+ "learning_rate": 8.937065143526346e-08,
7509
+ "logits/chosen": -1.2513972520828247,
7510
+ "logits/rejected": -1.2623106241226196,
7511
+ "logps/chosen": -151.0131378173828,
7512
+ "logps/rejected": -171.0509490966797,
7513
+ "loss": 0.4517,
7514
+ "rewards/accuracies": 0.6875,
7515
+ "rewards/chosen": -0.1952919065952301,
7516
+ "rewards/margins": 0.6836872696876526,
7517
+ "rewards/rejected": -0.8789791464805603,
7518
+ "step": 982
7519
+ },
7520
+ {
7521
+ "epoch": 1.1220868077827668,
7522
+ "grad_norm": 71.24214091954167,
7523
+ "learning_rate": 8.899254582510367e-08,
7524
+ "logits/chosen": -1.2960295677185059,
7525
+ "logits/rejected": -1.3374513387680054,
7526
+ "logps/chosen": -170.47344970703125,
7527
+ "logps/rejected": -197.69920349121094,
7528
+ "loss": 0.4205,
7529
+ "rewards/accuracies": 0.84375,
7530
+ "rewards/chosen": -0.2936790883541107,
7531
+ "rewards/margins": 1.0447689294815063,
7532
+ "rewards/rejected": -1.3384480476379395,
7533
+ "step": 984
7534
+ },
7535
+ {
7536
+ "epoch": 1.1243674720262276,
7537
+ "grad_norm": 58.68134074489735,
7538
+ "learning_rate": 8.861459944535389e-08,
7539
+ "logits/chosen": -1.3058708906173706,
7540
+ "logits/rejected": -1.3617157936096191,
7541
+ "logps/chosen": -149.42727661132812,
7542
+ "logps/rejected": -179.4463653564453,
7543
+ "loss": 0.4586,
7544
+ "rewards/accuracies": 0.75,
7545
+ "rewards/chosen": -0.1447882056236267,
7546
+ "rewards/margins": 0.7445257902145386,
7547
+ "rewards/rejected": -0.8893140554428101,
7548
+ "step": 986
7549
+ },
7550
+ {
7551
+ "epoch": 1.1266481362696885,
7552
+ "grad_norm": 61.951258926530045,
7553
+ "learning_rate": 8.823681776326898e-08,
7554
+ "logits/chosen": -1.3737095594406128,
7555
+ "logits/rejected": -1.4248648881912231,
7556
+ "logps/chosen": -189.31312561035156,
7557
+ "logps/rejected": -220.4053192138672,
7558
+ "loss": 0.4602,
7559
+ "rewards/accuracies": 0.8125,
7560
+ "rewards/chosen": -0.6107076406478882,
7561
+ "rewards/margins": 0.6425344944000244,
7562
+ "rewards/rejected": -1.2532421350479126,
7563
+ "step": 988
7564
+ },
7565
+ {
7566
+ "epoch": 1.1289288005131495,
7567
+ "grad_norm": 62.59059177140616,
7568
+ "learning_rate": 8.785920624372121e-08,
7569
+ "logits/chosen": -1.169434666633606,
7570
+ "logits/rejected": -1.1585787534713745,
7571
+ "logps/chosen": -154.62611389160156,
7572
+ "logps/rejected": -163.4484100341797,
7573
+ "loss": 0.4619,
7574
+ "rewards/accuracies": 0.75,
7575
+ "rewards/chosen": -0.1728251576423645,
7576
+ "rewards/margins": 0.8388264775276184,
7577
+ "rewards/rejected": -1.0116515159606934,
7578
+ "step": 990
7579
+ },
7580
+ {
7581
+ "epoch": 1.1312094647566104,
7582
+ "grad_norm": 56.29054558070934,
7583
+ "learning_rate": 8.748177034912147e-08,
7584
+ "logits/chosen": -1.4068464040756226,
7585
+ "logits/rejected": -1.4235914945602417,
7586
+ "logps/chosen": -203.57357788085938,
7587
+ "logps/rejected": -247.82298278808594,
7588
+ "loss": 0.4738,
7589
+ "rewards/accuracies": 0.84375,
7590
+ "rewards/chosen": -0.4565941393375397,
7591
+ "rewards/margins": 1.059171438217163,
7592
+ "rewards/rejected": -1.5157654285430908,
7593
+ "step": 992
7594
+ },
7595
+ {
7596
+ "epoch": 1.1334901290000712,
7597
+ "grad_norm": 51.085177964624826,
7598
+ "learning_rate": 8.710451553934008e-08,
7599
+ "logits/chosen": -1.3480877876281738,
7600
+ "logits/rejected": -1.3525364398956299,
7601
+ "logps/chosen": -155.8323974609375,
7602
+ "logps/rejected": -188.3721466064453,
7603
+ "loss": 0.4321,
7604
+ "rewards/accuracies": 0.8125,
7605
+ "rewards/chosen": -0.2992142140865326,
7606
+ "rewards/margins": 0.7550831437110901,
7607
+ "rewards/rejected": -1.0542973279953003,
7608
+ "step": 994
7609
+ },
7610
+ {
7611
+ "epoch": 1.1357707932435321,
7612
+ "grad_norm": 58.843609149415194,
7613
+ "learning_rate": 8.672744727162781e-08,
7614
+ "logits/chosen": -1.2812210321426392,
7615
+ "logits/rejected": -1.2667509317398071,
7616
+ "logps/chosen": -148.864990234375,
7617
+ "logps/rejected": -171.8406219482422,
7618
+ "loss": 0.4565,
7619
+ "rewards/accuracies": 0.78125,
7620
+ "rewards/chosen": -0.2008969783782959,
7621
+ "rewards/margins": 0.9470231533050537,
7622
+ "rewards/rejected": -1.1479202508926392,
7623
+ "step": 996
7624
+ },
7625
+ {
7626
+ "epoch": 1.138051457486993,
7627
+ "grad_norm": 55.1542603604775,
7628
+ "learning_rate": 8.635057100053701e-08,
7629
+ "logits/chosen": -1.2231156826019287,
7630
+ "logits/rejected": -1.2690283060073853,
7631
+ "logps/chosen": -134.86245727539062,
7632
+ "logps/rejected": -175.73106384277344,
7633
+ "loss": 0.4239,
7634
+ "rewards/accuracies": 0.9375,
7635
+ "rewards/chosen": -0.14157333970069885,
7636
+ "rewards/margins": 1.2773213386535645,
7637
+ "rewards/rejected": -1.4188947677612305,
7638
+ "step": 998
7639
+ },
7640
+ {
7641
+ "epoch": 1.140332121730454,
7642
+ "grad_norm": 61.40854796108108,
7643
+ "learning_rate": 8.597389217784267e-08,
7644
+ "logits/chosen": -1.2670400142669678,
7645
+ "logits/rejected": -1.3082315921783447,
7646
+ "logps/chosen": -159.52301025390625,
7647
+ "logps/rejected": -179.5859832763672,
7648
+ "loss": 0.4284,
7649
+ "rewards/accuracies": 0.875,
7650
+ "rewards/chosen": -0.2508872151374817,
7651
+ "rewards/margins": 0.8329470157623291,
7652
+ "rewards/rejected": -1.083834171295166,
7653
+ "step": 1000
7654
+ },
7655
+ {
7656
+ "epoch": 1.140332121730454,
7657
+ "eval_logits/chosen": -1.3981410264968872,
7658
+ "eval_logits/rejected": -1.382390022277832,
7659
+ "eval_logps/chosen": -131.1500701904297,
7660
+ "eval_logps/rejected": -136.96005249023438,
7661
+ "eval_loss": 0.5724203586578369,
7662
+ "eval_rewards/accuracies": 0.6800000071525574,
7663
+ "eval_rewards/chosen": -0.12521003186702728,
7664
+ "eval_rewards/margins": 0.36452141404151917,
7665
+ "eval_rewards/rejected": -0.48973140120506287,
7666
+ "eval_runtime": 19.6429,
7667
+ "eval_samples_per_second": 5.091,
7668
+ "eval_steps_per_second": 1.273,
7669
+ "step": 1000
7670
+ },
7671
+ {
7672
+ "epoch": 1.142612785973915,
7673
+ "grad_norm": 67.69710183071128,
7674
+ "learning_rate": 8.55974162524635e-08,
7675
+ "logits/chosen": -1.1661014556884766,
7676
+ "logits/rejected": -1.1639918088912964,
7677
+ "logps/chosen": -152.52914428710938,
7678
+ "logps/rejected": -175.15721130371094,
7679
+ "loss": 0.4415,
7680
+ "rewards/accuracies": 0.6875,
7681
+ "rewards/chosen": -0.44051098823547363,
7682
+ "rewards/margins": 0.7846835255622864,
7683
+ "rewards/rejected": -1.2251945734024048,
7684
+ "step": 1002
7685
+ },
7686
+ {
7687
+ "epoch": 1.1448934502173758,
7688
+ "grad_norm": 60.82110671671323,
7689
+ "learning_rate": 8.522114867038322e-08,
7690
+ "logits/chosen": -1.3812358379364014,
7691
+ "logits/rejected": -1.3619797229766846,
7692
+ "logps/chosen": -155.48292541503906,
7693
+ "logps/rejected": -154.6956024169922,
7694
+ "loss": 0.4675,
7695
+ "rewards/accuracies": 0.90625,
7696
+ "rewards/chosen": -0.1355004459619522,
7697
+ "rewards/margins": 0.798072099685669,
7698
+ "rewards/rejected": -0.93357253074646,
7699
+ "step": 1004
7700
+ },
7701
+ {
7702
+ "epoch": 1.1471741144608367,
7703
+ "grad_norm": 53.52801508976481,
7704
+ "learning_rate": 8.484509487457171e-08,
7705
+ "logits/chosen": -1.3384727239608765,
7706
+ "logits/rejected": -1.3696333169937134,
7707
+ "logps/chosen": -202.6947479248047,
7708
+ "logps/rejected": -222.9131622314453,
7709
+ "loss": 0.4287,
7710
+ "rewards/accuracies": 0.84375,
7711
+ "rewards/chosen": -0.21906551718711853,
7712
+ "rewards/margins": 1.0475835800170898,
7713
+ "rewards/rejected": -1.2666492462158203,
7714
+ "step": 1006
7715
+ },
7716
+ {
7717
+ "epoch": 1.1494547787042977,
7718
+ "grad_norm": 51.71321310860304,
7719
+ "learning_rate": 8.446926030490622e-08,
7720
+ "logits/chosen": -1.391283392906189,
7721
+ "logits/rejected": -1.4445586204528809,
7722
+ "logps/chosen": -123.97080993652344,
7723
+ "logps/rejected": -172.02490234375,
7724
+ "loss": 0.4404,
7725
+ "rewards/accuracies": 0.78125,
7726
+ "rewards/chosen": -0.22346752882003784,
7727
+ "rewards/margins": 0.7628045082092285,
7728
+ "rewards/rejected": -0.9862719774246216,
7729
+ "step": 1008
7730
+ },
7731
+ {
7732
+ "epoch": 1.1517354429477584,
7733
+ "grad_norm": 50.571283153367446,
7734
+ "learning_rate": 8.40936503980928e-08,
7735
+ "logits/chosen": -1.2062405347824097,
7736
+ "logits/rejected": -1.2253963947296143,
7737
+ "logps/chosen": -225.68972778320312,
7738
+ "logps/rejected": -253.89883422851562,
7739
+ "loss": 0.4194,
7740
+ "rewards/accuracies": 0.90625,
7741
+ "rewards/chosen": -0.2782267928123474,
7742
+ "rewards/margins": 1.586256504058838,
7743
+ "rewards/rejected": -1.86448335647583,
7744
+ "step": 1010
7745
+ },
7746
+ {
7747
+ "epoch": 1.1540161071912194,
7748
+ "grad_norm": 104.63385992757593,
7749
+ "learning_rate": 8.371827058758761e-08,
7750
+ "logits/chosen": -1.3567335605621338,
7751
+ "logits/rejected": -1.373883843421936,
7752
+ "logps/chosen": -173.80685424804688,
7753
+ "logps/rejected": -193.8018798828125,
7754
+ "loss": 0.4405,
7755
+ "rewards/accuracies": 0.875,
7756
+ "rewards/chosen": -0.4084673523902893,
7757
+ "rewards/margins": 0.7369635701179504,
7758
+ "rewards/rejected": -1.1454309225082397,
7759
+ "step": 1012
7760
+ },
7761
+ {
7762
+ "epoch": 1.1562967714346803,
7763
+ "grad_norm": 64.35691482136876,
7764
+ "learning_rate": 8.334312630351828e-08,
7765
+ "logits/chosen": -1.3794384002685547,
7766
+ "logits/rejected": -1.4179136753082275,
7767
+ "logps/chosen": -164.9583282470703,
7768
+ "logps/rejected": -195.81439208984375,
7769
+ "loss": 0.4849,
7770
+ "rewards/accuracies": 0.8125,
7771
+ "rewards/chosen": -0.21411646902561188,
7772
+ "rewards/margins": 0.8294214606285095,
7773
+ "rewards/rejected": -1.043537974357605,
7774
+ "step": 1014
7775
+ },
7776
+ {
7777
+ "epoch": 1.1585774356781413,
7778
+ "grad_norm": 55.14828540532963,
7779
+ "learning_rate": 8.29682229726054e-08,
7780
+ "logits/chosen": -1.2123723030090332,
7781
+ "logits/rejected": -1.2490003108978271,
7782
+ "logps/chosen": -170.7882537841797,
7783
+ "logps/rejected": -199.4991912841797,
7784
+ "loss": 0.4305,
7785
+ "rewards/accuracies": 0.78125,
7786
+ "rewards/chosen": -0.279404878616333,
7787
+ "rewards/margins": 0.8790395855903625,
7788
+ "rewards/rejected": -1.1584445238113403,
7789
+ "step": 1016
7790
+ },
7791
+ {
7792
+ "epoch": 1.1608580999216023,
7793
+ "grad_norm": 56.75948473097209,
7794
+ "learning_rate": 8.2593566018084e-08,
7795
+ "logits/chosen": -1.218788743019104,
7796
+ "logits/rejected": -1.3168693780899048,
7797
+ "logps/chosen": -198.46771240234375,
7798
+ "logps/rejected": -236.20420837402344,
7799
+ "loss": 0.3977,
7800
+ "rewards/accuracies": 0.875,
7801
+ "rewards/chosen": -0.30342423915863037,
7802
+ "rewards/margins": 1.051550269126892,
7803
+ "rewards/rejected": -1.3549745082855225,
7804
+ "step": 1018
7805
+ },
7806
+ {
7807
+ "epoch": 1.163138764165063,
7808
+ "grad_norm": 69.31406225752083,
7809
+ "learning_rate": 8.22191608596251e-08,
7810
+ "logits/chosen": -1.1068880558013916,
7811
+ "logits/rejected": -1.1753079891204834,
7812
+ "logps/chosen": -151.9117889404297,
7813
+ "logps/rejected": -186.4574432373047,
7814
+ "loss": 0.5023,
7815
+ "rewards/accuracies": 0.75,
7816
+ "rewards/chosen": -0.28111034631729126,
7817
+ "rewards/margins": 0.7174416184425354,
7818
+ "rewards/rejected": -0.9985520839691162,
7819
+ "step": 1020
7820
+ },
7821
+ {
7822
+ "epoch": 1.165419428408524,
7823
+ "grad_norm": 66.36797160504774,
7824
+ "learning_rate": 8.184501291325732e-08,
7825
+ "logits/chosen": -1.4255850315093994,
7826
+ "logits/rejected": -1.4551324844360352,
7827
+ "logps/chosen": -154.44932556152344,
7828
+ "logps/rejected": -180.50991821289062,
7829
+ "loss": 0.4782,
7830
+ "rewards/accuracies": 0.8125,
7831
+ "rewards/chosen": -0.34893348813056946,
7832
+ "rewards/margins": 0.6230674982070923,
7833
+ "rewards/rejected": -0.9720010757446289,
7834
+ "step": 1022
7835
+ },
7836
+ {
7837
+ "epoch": 1.167700092651985,
7838
+ "grad_norm": 69.9233343096583,
7839
+ "learning_rate": 8.147112759128858e-08,
7840
+ "logits/chosen": -1.2810206413269043,
7841
+ "logits/rejected": -1.3186777830123901,
7842
+ "logps/chosen": -195.02601623535156,
7843
+ "logps/rejected": -233.76730346679688,
7844
+ "loss": 0.4403,
7845
+ "rewards/accuracies": 0.875,
7846
+ "rewards/chosen": -0.3124176263809204,
7847
+ "rewards/margins": 1.22174870967865,
7848
+ "rewards/rejected": -1.5341663360595703,
7849
+ "step": 1024
7850
+ },
7851
+ {
7852
+ "epoch": 1.169980756895446,
7853
+ "grad_norm": 63.32204131506361,
7854
+ "learning_rate": 8.109751030222764e-08,
7855
+ "logits/chosen": -1.3566696643829346,
7856
+ "logits/rejected": -1.4292207956314087,
7857
+ "logps/chosen": -202.13467407226562,
7858
+ "logps/rejected": -226.0039825439453,
7859
+ "loss": 0.4815,
7860
+ "rewards/accuracies": 0.8125,
7861
+ "rewards/chosen": -0.3643057644367218,
7862
+ "rewards/margins": 0.7674057483673096,
7863
+ "rewards/rejected": -1.1317113637924194,
7864
+ "step": 1026
7865
+ },
7866
+ {
7867
+ "epoch": 1.1722614211389066,
7868
+ "grad_norm": 47.95687326599466,
7869
+ "learning_rate": 8.072416645070608e-08,
7870
+ "logits/chosen": -1.331808090209961,
7871
+ "logits/rejected": -1.375180959701538,
7872
+ "logps/chosen": -158.04380798339844,
7873
+ "logps/rejected": -180.04217529296875,
7874
+ "loss": 0.4145,
7875
+ "rewards/accuracies": 0.90625,
7876
+ "rewards/chosen": -0.10679011046886444,
7877
+ "rewards/margins": 0.9117978811264038,
7878
+ "rewards/rejected": -1.0185880661010742,
7879
+ "step": 1028
7880
+ },
7881
+ {
7882
+ "epoch": 1.1745420853823676,
7883
+ "grad_norm": 56.45833885104353,
7884
+ "learning_rate": 8.035110143739999e-08,
7885
+ "logits/chosen": -1.246973991394043,
7886
+ "logits/rejected": -1.2526628971099854,
7887
+ "logps/chosen": -149.625,
7888
+ "logps/rejected": -221.7806396484375,
7889
+ "loss": 0.458,
7890
+ "rewards/accuracies": 0.75,
7891
+ "rewards/chosen": -0.383423775434494,
7892
+ "rewards/margins": 0.8781226873397827,
7893
+ "rewards/rejected": -1.2615464925765991,
7894
+ "step": 1030
7895
+ },
7896
+ {
7897
+ "epoch": 1.1768227496258286,
7898
+ "grad_norm": 54.84274531559865,
7899
+ "learning_rate": 7.997832065895187e-08,
7900
+ "logits/chosen": -1.3046379089355469,
7901
+ "logits/rejected": -1.335304617881775,
7902
+ "logps/chosen": -184.51785278320312,
7903
+ "logps/rejected": -226.02542114257812,
7904
+ "loss": 0.4619,
7905
+ "rewards/accuracies": 0.78125,
7906
+ "rewards/chosen": -0.39717113971710205,
7907
+ "rewards/margins": 0.9696211218833923,
7908
+ "rewards/rejected": -1.3667922019958496,
7909
+ "step": 1032
7910
+ },
7911
+ {
7912
+ "epoch": 1.1791034138692895,
7913
+ "grad_norm": 70.40644247341677,
7914
+ "learning_rate": 7.960582950789257e-08,
7915
+ "logits/chosen": -1.4094276428222656,
7916
+ "logits/rejected": -1.3907825946807861,
7917
+ "logps/chosen": -177.15933227539062,
7918
+ "logps/rejected": -189.71926879882812,
7919
+ "loss": 0.466,
7920
+ "rewards/accuracies": 0.8125,
7921
+ "rewards/chosen": -0.43957361578941345,
7922
+ "rewards/margins": 0.5969427824020386,
7923
+ "rewards/rejected": -1.0365164279937744,
7924
+ "step": 1034
7925
+ },
7926
+ {
7927
+ "epoch": 1.1813840781127503,
7928
+ "grad_norm": 59.549418196573015,
7929
+ "learning_rate": 7.923363337256327e-08,
7930
+ "logits/chosen": -1.3654837608337402,
7931
+ "logits/rejected": -1.4075124263763428,
7932
+ "logps/chosen": -153.509033203125,
7933
+ "logps/rejected": -202.06356811523438,
7934
+ "loss": 0.4249,
7935
+ "rewards/accuracies": 0.90625,
7936
+ "rewards/chosen": -0.264252245426178,
7937
+ "rewards/margins": 1.4065436124801636,
7938
+ "rewards/rejected": -1.6707959175109863,
7939
+ "step": 1036
7940
+ },
7941
+ {
7942
+ "epoch": 1.1836647423562112,
7943
+ "grad_norm": 60.574371000814885,
7944
+ "learning_rate": 7.886173763703757e-08,
7945
+ "logits/chosen": -1.309372067451477,
7946
+ "logits/rejected": -1.3306725025177002,
7947
+ "logps/chosen": -147.5194091796875,
7948
+ "logps/rejected": -155.11155700683594,
7949
+ "loss": 0.4683,
7950
+ "rewards/accuracies": 0.625,
7951
+ "rewards/chosen": -0.18043869733810425,
7952
+ "rewards/margins": 0.39405256509780884,
7953
+ "rewards/rejected": -0.5744912624359131,
7954
+ "step": 1038
7955
+ },
7956
+ {
7957
+ "epoch": 1.1859454065996722,
7958
+ "grad_norm": 58.21498952759633,
7959
+ "learning_rate": 7.849014768104353e-08,
7960
+ "logits/chosen": -1.4345394372940063,
7961
+ "logits/rejected": -1.454176425933838,
7962
+ "logps/chosen": -208.239501953125,
7963
+ "logps/rejected": -231.63497924804688,
7964
+ "loss": 0.4582,
7965
+ "rewards/accuracies": 0.8125,
7966
+ "rewards/chosen": -0.41373205184936523,
7967
+ "rewards/margins": 0.9180522561073303,
7968
+ "rewards/rejected": -1.3317843675613403,
7969
+ "step": 1040
7970
+ },
7971
+ {
7972
+ "epoch": 1.1882260708431331,
7973
+ "grad_norm": 60.11382019448424,
7974
+ "learning_rate": 7.811886887988596e-08,
7975
+ "logits/chosen": -1.1493500471115112,
7976
+ "logits/rejected": -1.2578613758087158,
7977
+ "logps/chosen": -156.71392822265625,
7978
+ "logps/rejected": -203.9082794189453,
7979
+ "loss": 0.43,
7980
+ "rewards/accuracies": 0.78125,
7981
+ "rewards/chosen": -0.3507915437221527,
7982
+ "rewards/margins": 0.6405390501022339,
7983
+ "rewards/rejected": -0.991330623626709,
7984
+ "step": 1042
7985
+ },
7986
+ {
7987
+ "epoch": 1.1905067350865939,
7988
+ "grad_norm": 62.63091780254508,
7989
+ "learning_rate": 7.774790660436856e-08,
7990
+ "logits/chosen": -1.35098135471344,
7991
+ "logits/rejected": -1.3219125270843506,
7992
+ "logps/chosen": -164.2214813232422,
7993
+ "logps/rejected": -184.14907836914062,
7994
+ "loss": 0.4559,
7995
+ "rewards/accuracies": 0.78125,
7996
+ "rewards/chosen": -0.37730497121810913,
7997
+ "rewards/margins": 0.7604411244392395,
7998
+ "rewards/rejected": -1.1377460956573486,
7999
+ "step": 1044
8000
+ },
8001
+ {
8002
+ "epoch": 1.1927873993300548,
8003
+ "grad_norm": 58.393702096339915,
8004
+ "learning_rate": 7.737726622071628e-08,
8005
+ "logits/chosen": -1.2446062564849854,
8006
+ "logits/rejected": -1.2715435028076172,
8007
+ "logps/chosen": -161.71238708496094,
8008
+ "logps/rejected": -208.52255249023438,
8009
+ "loss": 0.4614,
8010
+ "rewards/accuracies": 0.71875,
8011
+ "rewards/chosen": -0.5151055455207825,
8012
+ "rewards/margins": 0.7842628359794617,
8013
+ "rewards/rejected": -1.2993682622909546,
8014
+ "step": 1046
8015
+ },
8016
+ {
8017
+ "epoch": 1.1950680635735158,
8018
+ "grad_norm": 68.88961801831904,
8019
+ "learning_rate": 7.700695309049766e-08,
8020
+ "logits/chosen": -1.3265782594680786,
8021
+ "logits/rejected": -1.3526149988174438,
8022
+ "logps/chosen": -155.804443359375,
8023
+ "logps/rejected": -158.39266967773438,
8024
+ "loss": 0.4413,
8025
+ "rewards/accuracies": 0.71875,
8026
+ "rewards/chosen": -0.2930993437767029,
8027
+ "rewards/margins": 0.6833476424217224,
8028
+ "rewards/rejected": -0.9764469861984253,
8029
+ "step": 1048
8030
+ },
8031
+ {
8032
+ "epoch": 1.1973487278169768,
8033
+ "grad_norm": 62.58520626376503,
8034
+ "learning_rate": 7.663697257054736e-08,
8035
+ "logits/chosen": -1.193183183670044,
8036
+ "logits/rejected": -1.3192574977874756,
8037
+ "logps/chosen": -146.06712341308594,
8038
+ "logps/rejected": -209.09132385253906,
8039
+ "loss": 0.4365,
8040
+ "rewards/accuracies": 0.90625,
8041
+ "rewards/chosen": -0.26093000173568726,
8042
+ "rewards/margins": 0.8910123705863953,
8043
+ "rewards/rejected": -1.151942253112793,
8044
+ "step": 1050
8045
+ },
8046
+ {
8047
+ "epoch": 1.1996293920604377,
8048
+ "grad_norm": 60.54325139288285,
8049
+ "learning_rate": 7.62673300128885e-08,
8050
+ "logits/chosen": -1.2697768211364746,
8051
+ "logits/rejected": -1.3025808334350586,
8052
+ "logps/chosen": -140.79417419433594,
8053
+ "logps/rejected": -190.2462158203125,
8054
+ "loss": 0.4533,
8055
+ "rewards/accuracies": 0.84375,
8056
+ "rewards/chosen": -0.18518134951591492,
8057
+ "rewards/margins": 0.8462077975273132,
8058
+ "rewards/rejected": -1.0313892364501953,
8059
+ "step": 1052
8060
+ },
8061
+ {
8062
+ "epoch": 1.2019100563038985,
8063
+ "grad_norm": 62.02420670554748,
8064
+ "learning_rate": 7.589803076465546e-08,
8065
+ "logits/chosen": -1.3338916301727295,
8066
+ "logits/rejected": -1.3496989011764526,
8067
+ "logps/chosen": -190.69876098632812,
8068
+ "logps/rejected": -227.00088500976562,
8069
+ "loss": 0.4498,
8070
+ "rewards/accuracies": 0.84375,
8071
+ "rewards/chosen": -0.321012020111084,
8072
+ "rewards/margins": 1.0334088802337646,
8073
+ "rewards/rejected": -1.3544210195541382,
8074
+ "step": 1054
8075
+ },
8076
+ {
8077
+ "epoch": 1.2041907205473594,
8078
+ "grad_norm": 65.72599471201299,
8079
+ "learning_rate": 7.552908016801632e-08,
8080
+ "logits/chosen": -1.3135321140289307,
8081
+ "logits/rejected": -1.3163644075393677,
8082
+ "logps/chosen": -183.17430114746094,
8083
+ "logps/rejected": -195.53515625,
8084
+ "loss": 0.4853,
8085
+ "rewards/accuracies": 0.78125,
8086
+ "rewards/chosen": -0.4703221619129181,
8087
+ "rewards/margins": 0.690421998500824,
8088
+ "rewards/rejected": -1.1607441902160645,
8089
+ "step": 1056
8090
+ },
8091
+ {
8092
+ "epoch": 1.2064713847908204,
8093
+ "grad_norm": 67.75915859922333,
8094
+ "learning_rate": 7.516048356009577e-08,
8095
+ "logits/chosen": -1.309540867805481,
8096
+ "logits/rejected": -1.3292198181152344,
8097
+ "logps/chosen": -169.94964599609375,
8098
+ "logps/rejected": -181.25917053222656,
8099
+ "loss": 0.4585,
8100
+ "rewards/accuracies": 0.75,
8101
+ "rewards/chosen": -0.44265216588974,
8102
+ "rewards/margins": 0.8768545985221863,
8103
+ "rewards/rejected": -1.3195068836212158,
8104
+ "step": 1058
8105
+ },
8106
+ {
8107
+ "epoch": 1.2087520490342811,
8108
+ "grad_norm": 60.73391494662122,
8109
+ "learning_rate": 7.479224627289764e-08,
8110
+ "logits/chosen": -1.2144145965576172,
8111
+ "logits/rejected": -1.2872897386550903,
8112
+ "logps/chosen": -129.9945831298828,
8113
+ "logps/rejected": -174.8819580078125,
8114
+ "loss": 0.4193,
8115
+ "rewards/accuracies": 0.84375,
8116
+ "rewards/chosen": -0.19057044386863708,
8117
+ "rewards/margins": 0.8820828199386597,
8118
+ "rewards/rejected": -1.0726532936096191,
8119
+ "step": 1060
8120
+ },
8121
+ {
8122
+ "epoch": 1.211032713277742,
8123
+ "grad_norm": 57.13712403831016,
8124
+ "learning_rate": 7.442437363322812e-08,
8125
+ "logits/chosen": -1.307482361793518,
8126
+ "logits/rejected": -1.3386861085891724,
8127
+ "logps/chosen": -186.2252197265625,
8128
+ "logps/rejected": -227.38156127929688,
8129
+ "loss": 0.4097,
8130
+ "rewards/accuracies": 0.8125,
8131
+ "rewards/chosen": -0.2698296904563904,
8132
+ "rewards/margins": 1.0196818113327026,
8133
+ "rewards/rejected": -1.2895114421844482,
8134
+ "step": 1062
8135
+ },
8136
+ {
8137
+ "epoch": 1.213313377521203,
8138
+ "grad_norm": 61.28403933125535,
8139
+ "learning_rate": 7.405687096261839e-08,
8140
+ "logits/chosen": -1.3965141773223877,
8141
+ "logits/rejected": -1.4495915174484253,
8142
+ "logps/chosen": -220.30531311035156,
8143
+ "logps/rejected": -261.9841613769531,
8144
+ "loss": 0.4575,
8145
+ "rewards/accuracies": 0.8125,
8146
+ "rewards/chosen": -0.45436590909957886,
8147
+ "rewards/margins": 1.2083145380020142,
8148
+ "rewards/rejected": -1.6626802682876587,
8149
+ "step": 1064
8150
+ },
8151
+ {
8152
+ "epoch": 1.215594041764664,
8153
+ "grad_norm": 66.4214015966849,
8154
+ "learning_rate": 7.368974357724789e-08,
8155
+ "logits/chosen": -1.2381188869476318,
8156
+ "logits/rejected": -1.2131423950195312,
8157
+ "logps/chosen": -164.27276611328125,
8158
+ "logps/rejected": -279.0275573730469,
8159
+ "loss": 0.3947,
8160
+ "rewards/accuracies": 0.875,
8161
+ "rewards/chosen": -0.26724374294281006,
8162
+ "rewards/margins": 1.546440601348877,
8163
+ "rewards/rejected": -1.8136844635009766,
8164
+ "step": 1066
8165
+ },
8166
+ {
8167
+ "epoch": 1.217874706008125,
8168
+ "grad_norm": 52.96231029305883,
8169
+ "learning_rate": 7.332299678786722e-08,
8170
+ "logits/chosen": -1.1868572235107422,
8171
+ "logits/rejected": -1.224455714225769,
8172
+ "logps/chosen": -165.09494018554688,
8173
+ "logps/rejected": -200.85031127929688,
8174
+ "loss": 0.4338,
8175
+ "rewards/accuracies": 0.78125,
8176
+ "rewards/chosen": -0.35685399174690247,
8177
+ "rewards/margins": 1.0505053997039795,
8178
+ "rewards/rejected": -1.4073594808578491,
8179
+ "step": 1068
8180
+ },
8181
+ {
8182
+ "epoch": 1.2201553702515857,
8183
+ "grad_norm": 65.66793449701784,
8184
+ "learning_rate": 7.295663589972138e-08,
8185
+ "logits/chosen": -1.4080750942230225,
8186
+ "logits/rejected": -1.40921151638031,
8187
+ "logps/chosen": -157.87884521484375,
8188
+ "logps/rejected": -179.00694274902344,
8189
+ "loss": 0.4529,
8190
+ "rewards/accuracies": 0.90625,
8191
+ "rewards/chosen": -0.19519281387329102,
8192
+ "rewards/margins": 0.7116326093673706,
8193
+ "rewards/rejected": -0.9068254828453064,
8194
+ "step": 1070
8195
+ },
8196
+ {
8197
+ "epoch": 1.2224360344950467,
8198
+ "grad_norm": 58.03621828629834,
8199
+ "learning_rate": 7.259066621247316e-08,
8200
+ "logits/chosen": -1.3837441205978394,
8201
+ "logits/rejected": -1.4137468338012695,
8202
+ "logps/chosen": -169.6446075439453,
8203
+ "logps/rejected": -189.69842529296875,
8204
+ "loss": 0.4533,
8205
+ "rewards/accuracies": 0.75,
8206
+ "rewards/chosen": -0.2820315361022949,
8207
+ "rewards/margins": 0.677888035774231,
8208
+ "rewards/rejected": -0.9599195718765259,
8209
+ "step": 1072
8210
+ },
8211
+ {
8212
+ "epoch": 1.2247166987385076,
8213
+ "grad_norm": 65.14721879160109,
8214
+ "learning_rate": 7.222509302012621e-08,
8215
+ "logits/chosen": -1.3673007488250732,
8216
+ "logits/rejected": -1.3410099744796753,
8217
+ "logps/chosen": -240.47653198242188,
8218
+ "logps/rejected": -257.4016418457031,
8219
+ "loss": 0.4147,
8220
+ "rewards/accuracies": 0.84375,
8221
+ "rewards/chosen": -0.28077253699302673,
8222
+ "rewards/margins": 0.9187532663345337,
8223
+ "rewards/rejected": -1.1995258331298828,
8224
+ "step": 1074
8225
+ },
8226
+ {
8227
+ "epoch": 1.2269973629819684,
8228
+ "grad_norm": 61.93965308054813,
8229
+ "learning_rate": 7.185992161094876e-08,
8230
+ "logits/chosen": -1.3297830820083618,
8231
+ "logits/rejected": -1.2881451845169067,
8232
+ "logps/chosen": -202.92349243164062,
8233
+ "logps/rejected": -220.54769897460938,
8234
+ "loss": 0.4607,
8235
+ "rewards/accuracies": 0.78125,
8236
+ "rewards/chosen": -0.2249366044998169,
8237
+ "rewards/margins": 1.0748928785324097,
8238
+ "rewards/rejected": -1.2998292446136475,
8239
+ "step": 1076
8240
+ },
8241
+ {
8242
+ "epoch": 1.2292780272254293,
8243
+ "grad_norm": 71.36489123402376,
8244
+ "learning_rate": 7.149515726739678e-08,
8245
+ "logits/chosen": -1.2684777975082397,
8246
+ "logits/rejected": -1.3346588611602783,
8247
+ "logps/chosen": -135.23745727539062,
8248
+ "logps/rejected": -170.05029296875,
8249
+ "loss": 0.5085,
8250
+ "rewards/accuracies": 0.84375,
8251
+ "rewards/chosen": -0.37416163086891174,
8252
+ "rewards/margins": 0.7284354567527771,
8253
+ "rewards/rejected": -1.1025971174240112,
8254
+ "step": 1078
8255
+ },
8256
+ {
8257
+ "epoch": 1.2315586914688903,
8258
+ "grad_norm": 63.847418083603706,
8259
+ "learning_rate": 7.113080526603792e-08,
8260
+ "logits/chosen": -1.3170560598373413,
8261
+ "logits/rejected": -1.3284251689910889,
8262
+ "logps/chosen": -226.8069305419922,
8263
+ "logps/rejected": -265.87310791015625,
8264
+ "loss": 0.3977,
8265
+ "rewards/accuracies": 0.78125,
8266
+ "rewards/chosen": -0.6735376715660095,
8267
+ "rewards/margins": 1.2206555604934692,
8268
+ "rewards/rejected": -1.894193172454834,
8269
+ "step": 1080
8270
+ },
8271
+ {
8272
+ "epoch": 1.2338393557123513,
8273
+ "grad_norm": 71.41962712931492,
8274
+ "learning_rate": 7.076687087747495e-08,
8275
+ "logits/chosen": -1.2745944261550903,
8276
+ "logits/rejected": -1.3628857135772705,
8277
+ "logps/chosen": -167.65524291992188,
8278
+ "logps/rejected": -192.4241180419922,
8279
+ "loss": 0.5152,
8280
+ "rewards/accuracies": 0.84375,
8281
+ "rewards/chosen": -0.3967045545578003,
8282
+ "rewards/margins": 0.7864773869514465,
8283
+ "rewards/rejected": -1.1831820011138916,
8284
+ "step": 1082
8285
+ },
8286
+ {
8287
+ "epoch": 1.2361200199558122,
8288
+ "grad_norm": 63.43060169338188,
8289
+ "learning_rate": 7.040335936626957e-08,
8290
+ "logits/chosen": -1.2247568368911743,
8291
+ "logits/rejected": -1.334959864616394,
8292
+ "logps/chosen": -154.92623901367188,
8293
+ "logps/rejected": -192.81825256347656,
8294
+ "loss": 0.3931,
8295
+ "rewards/accuracies": 0.90625,
8296
+ "rewards/chosen": -0.37357938289642334,
8297
+ "rewards/margins": 1.085322380065918,
8298
+ "rewards/rejected": -1.4589016437530518,
8299
+ "step": 1084
8300
+ },
8301
+ {
8302
+ "epoch": 1.238400684199273,
8303
+ "grad_norm": 57.073138047606875,
8304
+ "learning_rate": 7.004027599086634e-08,
8305
+ "logits/chosen": -1.2639906406402588,
8306
+ "logits/rejected": -1.3193964958190918,
8307
+ "logps/chosen": -210.07972717285156,
8308
+ "logps/rejected": -277.6387634277344,
8309
+ "loss": 0.4629,
8310
+ "rewards/accuracies": 0.84375,
8311
+ "rewards/chosen": -0.7190876007080078,
8312
+ "rewards/margins": 1.5234203338623047,
8313
+ "rewards/rejected": -2.2425079345703125,
8314
+ "step": 1086
8315
+ },
8316
+ {
8317
+ "epoch": 1.240681348442734,
8318
+ "grad_norm": 59.243249538144596,
8319
+ "learning_rate": 6.967762600351646e-08,
8320
+ "logits/chosen": -1.2877843379974365,
8321
+ "logits/rejected": -1.3444148302078247,
8322
+ "logps/chosen": -159.09422302246094,
8323
+ "logps/rejected": -178.9310760498047,
8324
+ "loss": 0.4381,
8325
+ "rewards/accuracies": 0.6875,
8326
+ "rewards/chosen": -0.4013168215751648,
8327
+ "rewards/margins": 0.7007678747177124,
8328
+ "rewards/rejected": -1.1020846366882324,
8329
+ "step": 1088
8330
+ },
8331
+ {
8332
+ "epoch": 1.2429620126861949,
8333
+ "grad_norm": 61.681333545711915,
8334
+ "learning_rate": 6.93154146502019e-08,
8335
+ "logits/chosen": -1.1838276386260986,
8336
+ "logits/rejected": -1.178552269935608,
8337
+ "logps/chosen": -132.74107360839844,
8338
+ "logps/rejected": -161.54119873046875,
8339
+ "loss": 0.453,
8340
+ "rewards/accuracies": 0.875,
8341
+ "rewards/chosen": -0.2650779187679291,
8342
+ "rewards/margins": 0.9995027184486389,
8343
+ "rewards/rejected": -1.2645806074142456,
8344
+ "step": 1090
8345
+ },
8346
+ {
8347
+ "epoch": 1.2452426769296558,
8348
+ "grad_norm": 52.74342769824309,
8349
+ "learning_rate": 6.895364717055948e-08,
8350
+ "logits/chosen": -1.2756752967834473,
8351
+ "logits/rejected": -1.339556336402893,
8352
+ "logps/chosen": -166.25335693359375,
8353
+ "logps/rejected": -215.28515625,
8354
+ "loss": 0.4548,
8355
+ "rewards/accuracies": 0.8125,
8356
+ "rewards/chosen": -0.27491295337677,
8357
+ "rewards/margins": 0.772299587726593,
8358
+ "rewards/rejected": -1.0472124814987183,
8359
+ "step": 1092
8360
+ },
8361
+ {
8362
+ "epoch": 1.2475233411731166,
8363
+ "grad_norm": 67.05885459459954,
8364
+ "learning_rate": 6.859232879780515e-08,
8365
+ "logits/chosen": -1.2031257152557373,
8366
+ "logits/rejected": -1.2640451192855835,
8367
+ "logps/chosen": -167.1966094970703,
8368
+ "logps/rejected": -168.45986938476562,
8369
+ "loss": 0.438,
8370
+ "rewards/accuracies": 0.75,
8371
+ "rewards/chosen": -0.5370901823043823,
8372
+ "rewards/margins": 0.6850543022155762,
8373
+ "rewards/rejected": -1.222144603729248,
8374
+ "step": 1094
8375
+ },
8376
+ {
8377
+ "epoch": 1.2498040054165775,
8378
+ "grad_norm": 61.68077906707433,
8379
+ "learning_rate": 6.823146475865807e-08,
8380
+ "logits/chosen": -1.3824557065963745,
8381
+ "logits/rejected": -1.4685417413711548,
8382
+ "logps/chosen": -170.927490234375,
8383
+ "logps/rejected": -193.9454803466797,
8384
+ "loss": 0.4349,
8385
+ "rewards/accuracies": 0.90625,
8386
+ "rewards/chosen": -0.21729141473770142,
8387
+ "rewards/margins": 0.8320575952529907,
8388
+ "rewards/rejected": -1.049349069595337,
8389
+ "step": 1096
8390
+ },
8391
+ {
8392
+ "epoch": 1.2520846696600385,
8393
+ "grad_norm": 60.117698245447464,
8394
+ "learning_rate": 6.787106027326529e-08,
8395
+ "logits/chosen": -1.2755475044250488,
8396
+ "logits/rejected": -1.3001313209533691,
8397
+ "logps/chosen": -176.72113037109375,
8398
+ "logps/rejected": -194.21383666992188,
8399
+ "loss": 0.4634,
8400
+ "rewards/accuracies": 0.71875,
8401
+ "rewards/chosen": -0.5146129131317139,
8402
+ "rewards/margins": 0.5713402628898621,
8403
+ "rewards/rejected": -1.0859532356262207,
8404
+ "step": 1098
8405
+ },
8406
+ {
8407
+ "epoch": 1.2543653339034995,
8408
+ "grad_norm": 60.13020970533813,
8409
+ "learning_rate": 6.751112055512604e-08,
8410
+ "logits/chosen": -1.3733175992965698,
8411
+ "logits/rejected": -1.4645140171051025,
8412
+ "logps/chosen": -153.76385498046875,
8413
+ "logps/rejected": -180.75132751464844,
8414
+ "loss": 0.4132,
8415
+ "rewards/accuracies": 0.71875,
8416
+ "rewards/chosen": -0.1185435876250267,
8417
+ "rewards/margins": 0.8378748297691345,
8418
+ "rewards/rejected": -0.9564183950424194,
8419
+ "step": 1100
8420
+ },
8421
+ {
8422
+ "epoch": 1.2543653339034995,
8423
+ "eval_logits/chosen": -1.3715007305145264,
8424
+ "eval_logits/rejected": -1.3544986248016357,
8425
+ "eval_logps/chosen": -131.8278350830078,
8426
+ "eval_logps/rejected": -137.9906005859375,
8427
+ "eval_loss": 0.5563453435897827,
8428
+ "eval_rewards/accuracies": 0.7599999904632568,
8429
+ "eval_rewards/chosen": -0.19298861920833588,
8430
+ "eval_rewards/margins": 0.3997974693775177,
8431
+ "eval_rewards/rejected": -0.5927860736846924,
8432
+ "eval_runtime": 19.8497,
8433
+ "eval_samples_per_second": 5.038,
8434
+ "eval_steps_per_second": 1.259,
8435
+ "step": 1100
8436
+ },
8437
+ {
8438
+ "epoch": 1.2566459981469604,
8439
+ "grad_norm": 51.569029451796766,
8440
+ "learning_rate": 6.715165081101641e-08,
8441
+ "logits/chosen": -1.2064361572265625,
8442
+ "logits/rejected": -1.268586277961731,
8443
+ "logps/chosen": -170.99136352539062,
8444
+ "logps/rejected": -214.76968383789062,
8445
+ "loss": 0.4277,
8446
+ "rewards/accuracies": 0.8125,
8447
+ "rewards/chosen": -0.3422897160053253,
8448
+ "rewards/margins": 1.1319069862365723,
8449
+ "rewards/rejected": -1.4741966724395752,
8450
+ "step": 1102
8451
+ },
8452
+ {
8453
+ "epoch": 1.2589266623904212,
8454
+ "grad_norm": 68.83152296238853,
8455
+ "learning_rate": 6.679265624091392e-08,
8456
+ "logits/chosen": -1.3187097311019897,
8457
+ "logits/rejected": -1.3244363069534302,
8458
+ "logps/chosen": -177.85157775878906,
8459
+ "logps/rejected": -204.6130828857422,
8460
+ "loss": 0.483,
8461
+ "rewards/accuracies": 0.75,
8462
+ "rewards/chosen": -0.32719144225120544,
8463
+ "rewards/margins": 1.074700951576233,
8464
+ "rewards/rejected": -1.4018925428390503,
8465
+ "step": 1104
8466
+ },
8467
+ {
8468
+ "epoch": 1.2612073266338821,
8469
+ "grad_norm": 66.02612056942255,
8470
+ "learning_rate": 6.643414203792243e-08,
8471
+ "logits/chosen": -1.229019045829773,
8472
+ "logits/rejected": -1.3234455585479736,
8473
+ "logps/chosen": -184.5186004638672,
8474
+ "logps/rejected": -208.52243041992188,
8475
+ "loss": 0.4241,
8476
+ "rewards/accuracies": 0.84375,
8477
+ "rewards/chosen": -0.3899713158607483,
8478
+ "rewards/margins": 1.090962290763855,
8479
+ "rewards/rejected": -1.4809335470199585,
8480
+ "step": 1106
8481
+ },
8482
+ {
8483
+ "epoch": 1.263487990877343,
8484
+ "grad_norm": 59.215567502448074,
8485
+ "learning_rate": 6.607611338819696e-08,
8486
+ "logits/chosen": -1.2996883392333984,
8487
+ "logits/rejected": -1.319692850112915,
8488
+ "logps/chosen": -146.01797485351562,
8489
+ "logps/rejected": -178.61734008789062,
8490
+ "loss": 0.4377,
8491
+ "rewards/accuracies": 0.9375,
8492
+ "rewards/chosen": -0.31348976492881775,
8493
+ "rewards/margins": 0.9113309383392334,
8494
+ "rewards/rejected": -1.2248207330703735,
8495
+ "step": 1108
8496
+ },
8497
+ {
8498
+ "epoch": 1.2657686551208038,
8499
+ "grad_norm": 57.07885969385745,
8500
+ "learning_rate": 6.571857547086863e-08,
8501
+ "logits/chosen": -1.3365250825881958,
8502
+ "logits/rejected": -1.213597297668457,
8503
+ "logps/chosen": -202.40350341796875,
8504
+ "logps/rejected": -213.51048278808594,
8505
+ "loss": 0.4194,
8506
+ "rewards/accuracies": 0.875,
8507
+ "rewards/chosen": -0.38011184334754944,
8508
+ "rewards/margins": 1.0593245029449463,
8509
+ "rewards/rejected": -1.4394363164901733,
8510
+ "step": 1110
8511
+ },
8512
+ {
8513
+ "epoch": 1.2680493193642648,
8514
+ "grad_norm": 52.36073610875736,
8515
+ "learning_rate": 6.536153345796983e-08,
8516
+ "logits/chosen": -1.212735652923584,
8517
+ "logits/rejected": -1.2983253002166748,
8518
+ "logps/chosen": -208.72747802734375,
8519
+ "logps/rejected": -259.2344665527344,
8520
+ "loss": 0.4074,
8521
+ "rewards/accuracies": 0.8125,
8522
+ "rewards/chosen": -0.7484298348426819,
8523
+ "rewards/margins": 1.3461904525756836,
8524
+ "rewards/rejected": -2.0946204662323,
8525
+ "step": 1112
8526
+ },
8527
+ {
8528
+ "epoch": 1.2703299836077258,
8529
+ "grad_norm": 52.44708462406558,
8530
+ "learning_rate": 6.500499251435926e-08,
8531
+ "logits/chosen": -1.2901079654693604,
8532
+ "logits/rejected": -1.3678314685821533,
8533
+ "logps/chosen": -168.54234313964844,
8534
+ "logps/rejected": -214.02752685546875,
8535
+ "loss": 0.3998,
8536
+ "rewards/accuracies": 0.90625,
8537
+ "rewards/chosen": -0.32987186312675476,
8538
+ "rewards/margins": 1.0558862686157227,
8539
+ "rewards/rejected": -1.3857581615447998,
8540
+ "step": 1114
8541
+ },
8542
+ {
8543
+ "epoch": 1.2726106478511867,
8544
+ "grad_norm": 52.22832210272109,
8545
+ "learning_rate": 6.464895779764741e-08,
8546
+ "logits/chosen": -1.1556543111801147,
8547
+ "logits/rejected": -1.271488070487976,
8548
+ "logps/chosen": -131.3715057373047,
8549
+ "logps/rejected": -171.28237915039062,
8550
+ "loss": 0.4324,
8551
+ "rewards/accuracies": 0.90625,
8552
+ "rewards/chosen": -0.349406898021698,
8553
+ "rewards/margins": 0.989849328994751,
8554
+ "rewards/rejected": -1.3392562866210938,
8555
+ "step": 1116
8556
+ },
8557
+ {
8558
+ "epoch": 1.2748913120946477,
8559
+ "grad_norm": 62.70237944614458,
8560
+ "learning_rate": 6.429343445812177e-08,
8561
+ "logits/chosen": -1.146794319152832,
8562
+ "logits/rejected": -1.218197226524353,
8563
+ "logps/chosen": -194.02381896972656,
8564
+ "logps/rejected": -233.72328186035156,
8565
+ "loss": 0.4746,
8566
+ "rewards/accuracies": 0.75,
8567
+ "rewards/chosen": -0.5142129063606262,
8568
+ "rewards/margins": 0.9550236463546753,
8569
+ "rewards/rejected": -1.4692366123199463,
8570
+ "step": 1118
8571
+ },
8572
+ {
8573
+ "epoch": 1.2771719763381084,
8574
+ "grad_norm": 62.09523040000935,
8575
+ "learning_rate": 6.393842763867248e-08,
8576
+ "logits/chosen": -1.3159422874450684,
8577
+ "logits/rejected": -1.3710412979125977,
8578
+ "logps/chosen": -217.96322631835938,
8579
+ "logps/rejected": -217.71226501464844,
8580
+ "loss": 0.4275,
8581
+ "rewards/accuracies": 0.8125,
8582
+ "rewards/chosen": -0.5582747459411621,
8583
+ "rewards/margins": 0.7185696959495544,
8584
+ "rewards/rejected": -1.2768443822860718,
8585
+ "step": 1120
8586
+ },
8587
+ {
8588
+ "epoch": 1.2794526405815694,
8589
+ "grad_norm": 63.94486693511555,
8590
+ "learning_rate": 6.358394247471777e-08,
8591
+ "logits/chosen": -1.1774568557739258,
8592
+ "logits/rejected": -1.2224806547164917,
8593
+ "logps/chosen": -190.8328857421875,
8594
+ "logps/rejected": -234.33628845214844,
8595
+ "loss": 0.4315,
8596
+ "rewards/accuracies": 0.875,
8597
+ "rewards/chosen": -0.6915558576583862,
8598
+ "rewards/margins": 1.1698423624038696,
8599
+ "rewards/rejected": -1.8613982200622559,
8600
+ "step": 1122
8601
+ },
8602
+ {
8603
+ "epoch": 1.2817333048250303,
8604
+ "grad_norm": 59.101218821423586,
8605
+ "learning_rate": 6.322998409412988e-08,
8606
+ "logits/chosen": -1.33734929561615,
8607
+ "logits/rejected": -1.409659504890442,
8608
+ "logps/chosen": -187.33851623535156,
8609
+ "logps/rejected": -218.93324279785156,
8610
+ "loss": 0.439,
8611
+ "rewards/accuracies": 0.90625,
8612
+ "rewards/chosen": -0.5945166945457458,
8613
+ "rewards/margins": 0.9995917081832886,
8614
+ "rewards/rejected": -1.5941084623336792,
8615
+ "step": 1124
8616
+ },
8617
+ {
8618
+ "epoch": 1.284013969068491,
8619
+ "grad_norm": 59.13516077796158,
8620
+ "learning_rate": 6.287655761716067e-08,
8621
+ "logits/chosen": -1.3064815998077393,
8622
+ "logits/rejected": -1.3387144804000854,
8623
+ "logps/chosen": -191.6893768310547,
8624
+ "logps/rejected": -218.80178833007812,
8625
+ "loss": 0.4099,
8626
+ "rewards/accuracies": 0.90625,
8627
+ "rewards/chosen": -0.41915464401245117,
8628
+ "rewards/margins": 0.9991384744644165,
8629
+ "rewards/rejected": -1.4182932376861572,
8630
+ "step": 1126
8631
+ },
8632
+ {
8633
+ "epoch": 1.286294633311952,
8634
+ "grad_norm": 73.05044565719012,
8635
+ "learning_rate": 6.252366815636767e-08,
8636
+ "logits/chosen": -1.2290081977844238,
8637
+ "logits/rejected": -1.2006083726882935,
8638
+ "logps/chosen": -205.61956787109375,
8639
+ "logps/rejected": -196.34405517578125,
8640
+ "loss": 0.4412,
8641
+ "rewards/accuracies": 0.84375,
8642
+ "rewards/chosen": -0.7214517593383789,
8643
+ "rewards/margins": 0.7542897462844849,
8644
+ "rewards/rejected": -1.4757413864135742,
8645
+ "step": 1128
8646
+ },
8647
+ {
8648
+ "epoch": 1.288575297555413,
8649
+ "grad_norm": 58.9381764531572,
8650
+ "learning_rate": 6.217132081654014e-08,
8651
+ "logits/chosen": -1.2354235649108887,
8652
+ "logits/rejected": -1.2623546123504639,
8653
+ "logps/chosen": -160.61277770996094,
8654
+ "logps/rejected": -189.03482055664062,
8655
+ "loss": 0.4053,
8656
+ "rewards/accuracies": 0.78125,
8657
+ "rewards/chosen": -0.610559344291687,
8658
+ "rewards/margins": 1.2747613191604614,
8659
+ "rewards/rejected": -1.8853206634521484,
8660
+ "step": 1130
8661
+ },
8662
+ {
8663
+ "epoch": 1.290855961798874,
8664
+ "grad_norm": 63.57271664339758,
8665
+ "learning_rate": 6.181952069462513e-08,
8666
+ "logits/chosen": -1.2558073997497559,
8667
+ "logits/rejected": -1.3187260627746582,
8668
+ "logps/chosen": -182.9962158203125,
8669
+ "logps/rejected": -224.17080688476562,
8670
+ "loss": 0.4249,
8671
+ "rewards/accuracies": 0.78125,
8672
+ "rewards/chosen": -0.39853131771087646,
8673
+ "rewards/margins": 1.0594068765640259,
8674
+ "rewards/rejected": -1.4579381942749023,
8675
+ "step": 1132
8676
+ },
8677
+ {
8678
+ "epoch": 1.293136626042335,
8679
+ "grad_norm": 54.23291951216934,
8680
+ "learning_rate": 6.146827287965381e-08,
8681
+ "logits/chosen": -1.3496458530426025,
8682
+ "logits/rejected": -1.3700135946273804,
8683
+ "logps/chosen": -213.7442626953125,
8684
+ "logps/rejected": -243.01422119140625,
8685
+ "loss": 0.4322,
8686
+ "rewards/accuracies": 0.625,
8687
+ "rewards/chosen": -0.5680805444717407,
8688
+ "rewards/margins": 1.0590269565582275,
8689
+ "rewards/rejected": -1.6271076202392578,
8690
+ "step": 1134
8691
+ },
8692
+ {
8693
+ "epoch": 1.2954172902857957,
8694
+ "grad_norm": 56.45605987126908,
8695
+ "learning_rate": 6.111758245266794e-08,
8696
+ "logits/chosen": -1.3360776901245117,
8697
+ "logits/rejected": -1.2823421955108643,
8698
+ "logps/chosen": -135.13218688964844,
8699
+ "logps/rejected": -145.7807159423828,
8700
+ "loss": 0.4184,
8701
+ "rewards/accuracies": 0.75,
8702
+ "rewards/chosen": -0.2994767427444458,
8703
+ "rewards/margins": 0.7106032967567444,
8704
+ "rewards/rejected": -1.010080099105835,
8705
+ "step": 1136
8706
+ },
8707
+ {
8708
+ "epoch": 1.2976979545292566,
8709
+ "grad_norm": 67.42621295544383,
8710
+ "learning_rate": 6.076745448664615e-08,
8711
+ "logits/chosen": -1.3192064762115479,
8712
+ "logits/rejected": -1.3712198734283447,
8713
+ "logps/chosen": -173.24655151367188,
8714
+ "logps/rejected": -196.41253662109375,
8715
+ "loss": 0.4958,
8716
+ "rewards/accuracies": 0.84375,
8717
+ "rewards/chosen": -0.2075835019350052,
8718
+ "rewards/margins": 0.8700404763221741,
8719
+ "rewards/rejected": -1.077623963356018,
8720
+ "step": 1138
8721
+ },
8722
+ {
8723
+ "epoch": 1.2999786187727176,
8724
+ "grad_norm": 63.99357195093472,
8725
+ "learning_rate": 6.041789404643078e-08,
8726
+ "logits/chosen": -1.245415210723877,
8727
+ "logits/rejected": -1.3109159469604492,
8728
+ "logps/chosen": -131.79176330566406,
8729
+ "logps/rejected": -169.92637634277344,
8730
+ "loss": 0.4754,
8731
+ "rewards/accuracies": 0.75,
8732
+ "rewards/chosen": -0.31556999683380127,
8733
+ "rewards/margins": 0.7632789015769958,
8734
+ "rewards/rejected": -1.0788488388061523,
8735
+ "step": 1140
8736
+ },
8737
+ {
8738
+ "epoch": 1.3022592830161783,
8739
+ "grad_norm": 67.11890446031333,
8740
+ "learning_rate": 6.006890618865446e-08,
8741
+ "logits/chosen": -1.250425100326538,
8742
+ "logits/rejected": -1.3222934007644653,
8743
+ "logps/chosen": -192.59100341796875,
8744
+ "logps/rejected": -223.69459533691406,
8745
+ "loss": 0.4643,
8746
+ "rewards/accuracies": 0.875,
8747
+ "rewards/chosen": -0.32158908247947693,
8748
+ "rewards/margins": 1.268847107887268,
8749
+ "rewards/rejected": -1.5904361009597778,
8750
+ "step": 1142
8751
+ },
8752
+ {
8753
+ "epoch": 1.3045399472596393,
8754
+ "grad_norm": 76.00247534873895,
8755
+ "learning_rate": 5.972049596166705e-08,
8756
+ "logits/chosen": -1.1002310514450073,
8757
+ "logits/rejected": -1.1576155424118042,
8758
+ "logps/chosen": -148.41189575195312,
8759
+ "logps/rejected": -185.99664306640625,
8760
+ "loss": 0.4429,
8761
+ "rewards/accuracies": 0.875,
8762
+ "rewards/chosen": -0.632346510887146,
8763
+ "rewards/margins": 1.0468285083770752,
8764
+ "rewards/rejected": -1.6791750192642212,
8765
+ "step": 1144
8766
+ },
8767
+ {
8768
+ "epoch": 1.3068206115031002,
8769
+ "grad_norm": 70.43915324421563,
8770
+ "learning_rate": 5.9372668405462646e-08,
8771
+ "logits/chosen": -1.3556523323059082,
8772
+ "logits/rejected": -1.4173972606658936,
8773
+ "logps/chosen": -212.95169067382812,
8774
+ "logps/rejected": -238.5531768798828,
8775
+ "loss": 0.4643,
8776
+ "rewards/accuracies": 0.9375,
8777
+ "rewards/chosen": -0.40602514147758484,
8778
+ "rewards/margins": 1.0105608701705933,
8779
+ "rewards/rejected": -1.4165860414505005,
8780
+ "step": 1146
8781
+ },
8782
+ {
8783
+ "epoch": 1.3091012757465612,
8784
+ "grad_norm": 63.83492445068513,
8785
+ "learning_rate": 5.9025428551606415e-08,
8786
+ "logits/chosen": -1.244753360748291,
8787
+ "logits/rejected": -1.273009181022644,
8788
+ "logps/chosen": -164.7984619140625,
8789
+ "logps/rejected": -216.9807586669922,
8790
+ "loss": 0.4013,
8791
+ "rewards/accuracies": 0.90625,
8792
+ "rewards/chosen": -0.4480228126049042,
8793
+ "rewards/margins": 1.1551215648651123,
8794
+ "rewards/rejected": -1.6031445264816284,
8795
+ "step": 1148
8796
+ },
8797
+ {
8798
+ "epoch": 1.3113819399900222,
8799
+ "grad_norm": 60.05293035275106,
8800
+ "learning_rate": 5.867878142316221e-08,
8801
+ "logits/chosen": -1.3057725429534912,
8802
+ "logits/rejected": -1.3353570699691772,
8803
+ "logps/chosen": -145.6079559326172,
8804
+ "logps/rejected": -162.2236328125,
8805
+ "loss": 0.4074,
8806
+ "rewards/accuracies": 0.8125,
8807
+ "rewards/chosen": -0.13641442358493805,
8808
+ "rewards/margins": 0.7820553779602051,
8809
+ "rewards/rejected": -0.9184697270393372,
8810
+ "step": 1150
8811
+ },
8812
+ {
8813
+ "epoch": 1.3136626042334831,
8814
+ "grad_norm": 65.80534748016166,
8815
+ "learning_rate": 5.833273203461959e-08,
8816
+ "logits/chosen": -1.2780495882034302,
8817
+ "logits/rejected": -1.3019449710845947,
8818
+ "logps/chosen": -196.4916229248047,
8819
+ "logps/rejected": -222.07977294921875,
8820
+ "loss": 0.4397,
8821
+ "rewards/accuracies": 0.8125,
8822
+ "rewards/chosen": -0.3648252487182617,
8823
+ "rewards/margins": 1.3595582246780396,
8824
+ "rewards/rejected": -1.7243834733963013,
8825
+ "step": 1152
8826
+ },
8827
+ {
8828
+ "epoch": 1.3159432684769439,
8829
+ "grad_norm": 77.90046778105933,
8830
+ "learning_rate": 5.798728539182143e-08,
8831
+ "logits/chosen": -1.3802424669265747,
8832
+ "logits/rejected": -1.3905105590820312,
8833
+ "logps/chosen": -158.01589965820312,
8834
+ "logps/rejected": -197.33522033691406,
8835
+ "loss": 0.4201,
8836
+ "rewards/accuracies": 0.78125,
8837
+ "rewards/chosen": -0.4259686768054962,
8838
+ "rewards/margins": 0.7603559494018555,
8839
+ "rewards/rejected": -1.1863245964050293,
8840
+ "step": 1154
8841
+ },
8842
+ {
8843
+ "epoch": 1.3182239327204048,
8844
+ "grad_norm": 57.72791245543785,
8845
+ "learning_rate": 5.764244649189149e-08,
8846
+ "logits/chosen": -1.2129665613174438,
8847
+ "logits/rejected": -1.2697325944900513,
8848
+ "logps/chosen": -178.7556610107422,
8849
+ "logps/rejected": -201.34042358398438,
8850
+ "loss": 0.4094,
8851
+ "rewards/accuracies": 0.8125,
8852
+ "rewards/chosen": -0.3163738548755646,
8853
+ "rewards/margins": 0.971676766872406,
8854
+ "rewards/rejected": -1.288050651550293,
8855
+ "step": 1156
8856
+ },
8857
+ {
8858
+ "epoch": 1.3205045969638658,
8859
+ "grad_norm": 70.16500681265146,
8860
+ "learning_rate": 5.729822032316208e-08,
8861
+ "logits/chosen": -1.2184062004089355,
8862
+ "logits/rejected": -1.3518949747085571,
8863
+ "logps/chosen": -127.61387634277344,
8864
+ "logps/rejected": -177.82791137695312,
8865
+ "loss": 0.4509,
8866
+ "rewards/accuracies": 0.84375,
8867
+ "rewards/chosen": -0.3493412733078003,
8868
+ "rewards/margins": 1.1434459686279297,
8869
+ "rewards/rejected": -1.49278724193573,
8870
+ "step": 1158
8871
+ },
8872
+ {
8873
+ "epoch": 1.3227852612073265,
8874
+ "grad_norm": 65.43520672265807,
8875
+ "learning_rate": 5.6954611865101935e-08,
8876
+ "logits/chosen": -1.35573410987854,
8877
+ "logits/rejected": -1.4151753187179565,
8878
+ "logps/chosen": -219.81471252441406,
8879
+ "logps/rejected": -242.21942138671875,
8880
+ "loss": 0.4384,
8881
+ "rewards/accuracies": 0.90625,
8882
+ "rewards/chosen": -0.508373498916626,
8883
+ "rewards/margins": 1.2028579711914062,
8884
+ "rewards/rejected": -1.7112314701080322,
8885
+ "step": 1160
8886
+ },
8887
+ {
8888
+ "epoch": 1.3250659254507875,
8889
+ "grad_norm": 63.06305687675219,
8890
+ "learning_rate": 5.661162608824419e-08,
8891
+ "logits/chosen": -1.2267005443572998,
8892
+ "logits/rejected": -1.3211195468902588,
8893
+ "logps/chosen": -124.29898071289062,
8894
+ "logps/rejected": -159.11904907226562,
8895
+ "loss": 0.448,
8896
+ "rewards/accuracies": 0.90625,
8897
+ "rewards/chosen": -0.21557267010211945,
8898
+ "rewards/margins": 0.8949982523918152,
8899
+ "rewards/rejected": -1.110571026802063,
8900
+ "step": 1162
8901
+ },
8902
+ {
8903
+ "epoch": 1.3273465896942485,
8904
+ "grad_norm": 50.82426863386907,
8905
+ "learning_rate": 5.6269267954114464e-08,
8906
+ "logits/chosen": -1.232271671295166,
8907
+ "logits/rejected": -1.1851279735565186,
8908
+ "logps/chosen": -157.5489501953125,
8909
+ "logps/rejected": -258.4756774902344,
8910
+ "loss": 0.4687,
8911
+ "rewards/accuracies": 0.84375,
8912
+ "rewards/chosen": -0.4816133379936218,
8913
+ "rewards/margins": 1.42643141746521,
8914
+ "rewards/rejected": -1.9080448150634766,
8915
+ "step": 1164
8916
+ },
8917
+ {
8918
+ "epoch": 1.3296272539377094,
8919
+ "grad_norm": 59.70507538342753,
8920
+ "learning_rate": 5.592754241515908e-08,
8921
+ "logits/chosen": -1.3426597118377686,
8922
+ "logits/rejected": -1.379440426826477,
8923
+ "logps/chosen": -213.9489288330078,
8924
+ "logps/rejected": -245.739501953125,
8925
+ "loss": 0.4244,
8926
+ "rewards/accuracies": 0.8125,
8927
+ "rewards/chosen": -0.4688984155654907,
8928
+ "rewards/margins": 1.3086878061294556,
8929
+ "rewards/rejected": -1.7775862216949463,
8930
+ "step": 1166
8931
+ },
8932
+ {
8933
+ "epoch": 1.3319079181811704,
8934
+ "grad_norm": 68.73916859324237,
8935
+ "learning_rate": 5.5586454414673454e-08,
8936
+ "logits/chosen": -1.3552615642547607,
8937
+ "logits/rejected": -1.3137784004211426,
8938
+ "logps/chosen": -172.09751892089844,
8939
+ "logps/rejected": -183.2836456298828,
8940
+ "loss": 0.4544,
8941
+ "rewards/accuracies": 0.84375,
8942
+ "rewards/chosen": -0.30267298221588135,
8943
+ "rewards/margins": 1.0140584707260132,
8944
+ "rewards/rejected": -1.3167314529418945,
8945
+ "step": 1168
8946
+ },
8947
+ {
8948
+ "epoch": 1.3341885824246311,
8949
+ "grad_norm": 60.83087482532621,
8950
+ "learning_rate": 5.524600888673057e-08,
8951
+ "logits/chosen": -1.2173874378204346,
8952
+ "logits/rejected": -1.325149655342102,
8953
+ "logps/chosen": -156.27577209472656,
8954
+ "logps/rejected": -196.09017944335938,
8955
+ "loss": 0.4569,
8956
+ "rewards/accuracies": 0.9375,
8957
+ "rewards/chosen": -0.2902172803878784,
8958
+ "rewards/margins": 1.0630940198898315,
8959
+ "rewards/rejected": -1.3533114194869995,
8960
+ "step": 1170
8961
+ },
8962
+ {
8963
+ "epoch": 1.336469246668092,
8964
+ "grad_norm": 65.80542191040618,
8965
+ "learning_rate": 5.4906210756109575e-08,
8966
+ "logits/chosen": -1.223040223121643,
8967
+ "logits/rejected": -1.1960320472717285,
8968
+ "logps/chosen": -200.8917694091797,
8969
+ "logps/rejected": -229.284423828125,
8970
+ "loss": 0.453,
8971
+ "rewards/accuracies": 0.75,
8972
+ "rewards/chosen": -0.5325464606285095,
8973
+ "rewards/margins": 1.22454833984375,
8974
+ "rewards/rejected": -1.7570948600769043,
8975
+ "step": 1172
8976
+ },
8977
+ {
8978
+ "epoch": 1.338749910911553,
8979
+ "grad_norm": 60.664933897307115,
8980
+ "learning_rate": 5.456706493822458e-08,
8981
+ "logits/chosen": -1.3014591932296753,
8982
+ "logits/rejected": -1.277079463005066,
8983
+ "logps/chosen": -167.12942504882812,
8984
+ "logps/rejected": -180.93209838867188,
8985
+ "loss": 0.4185,
8986
+ "rewards/accuracies": 0.84375,
8987
+ "rewards/chosen": -0.23974554240703583,
8988
+ "rewards/margins": 0.8305579423904419,
8989
+ "rewards/rejected": -1.0703034400939941,
8990
+ "step": 1174
8991
+ },
8992
+ {
8993
+ "epoch": 1.3410305751550138,
8994
+ "grad_norm": 67.3008532319244,
8995
+ "learning_rate": 5.4228576339053577e-08,
8996
+ "logits/chosen": -1.2223035097122192,
8997
+ "logits/rejected": -1.2677267789840698,
8998
+ "logps/chosen": -170.18605041503906,
8999
+ "logps/rejected": -188.31619262695312,
9000
+ "loss": 0.4476,
9001
+ "rewards/accuracies": 0.75,
9002
+ "rewards/chosen": -0.4260440766811371,
9003
+ "rewards/margins": 0.6966915130615234,
9004
+ "rewards/rejected": -1.1227355003356934,
9005
+ "step": 1176
9006
+ },
9007
+ {
9008
+ "epoch": 1.3433112393984747,
9009
+ "grad_norm": 58.74035582903428,
9010
+ "learning_rate": 5.38907498550674e-08,
9011
+ "logits/chosen": -1.263763666152954,
9012
+ "logits/rejected": -1.3518104553222656,
9013
+ "logps/chosen": -218.0510711669922,
9014
+ "logps/rejected": -249.43292236328125,
9015
+ "loss": 0.4739,
9016
+ "rewards/accuracies": 0.96875,
9017
+ "rewards/chosen": -0.3573762774467468,
9018
+ "rewards/margins": 1.326561450958252,
9019
+ "rewards/rejected": -1.6839375495910645,
9020
+ "step": 1178
9021
+ },
9022
+ {
9023
+ "epoch": 1.3455919036419357,
9024
+ "grad_norm": 57.92848976074756,
9025
+ "learning_rate": 5.355359037315893e-08,
9026
+ "logits/chosen": -1.2013936042785645,
9027
+ "logits/rejected": -1.3312106132507324,
9028
+ "logps/chosen": -142.94082641601562,
9029
+ "logps/rejected": -221.46524047851562,
9030
+ "loss": 0.4267,
9031
+ "rewards/accuracies": 0.875,
9032
+ "rewards/chosen": -0.3215872645378113,
9033
+ "rewards/margins": 1.4045162200927734,
9034
+ "rewards/rejected": -1.726103663444519,
9035
+ "step": 1180
9036
+ },
9037
+ {
9038
+ "epoch": 1.3478725678853967,
9039
+ "grad_norm": 55.43645182737105,
9040
+ "learning_rate": 5.3217102770572466e-08,
9041
+ "logits/chosen": -1.3156689405441284,
9042
+ "logits/rejected": -1.2861976623535156,
9043
+ "logps/chosen": -195.7543182373047,
9044
+ "logps/rejected": -248.62838745117188,
9045
+ "loss": 0.4405,
9046
+ "rewards/accuracies": 0.84375,
9047
+ "rewards/chosen": -0.4373939037322998,
9048
+ "rewards/margins": 1.2866684198379517,
9049
+ "rewards/rejected": -1.7240623235702515,
9050
+ "step": 1182
9051
+ },
9052
+ {
9053
+ "epoch": 1.3501532321288576,
9054
+ "grad_norm": 57.579451083148605,
9055
+ "learning_rate": 5.288129191483296e-08,
9056
+ "logits/chosen": -1.2084459066390991,
9057
+ "logits/rejected": -1.1476478576660156,
9058
+ "logps/chosen": -136.50244140625,
9059
+ "logps/rejected": -144.5672149658203,
9060
+ "loss": 0.4725,
9061
+ "rewards/accuracies": 0.6875,
9062
+ "rewards/chosen": -0.35504642128944397,
9063
+ "rewards/margins": 0.5072875022888184,
9064
+ "rewards/rejected": -0.8623339533805847,
9065
+ "step": 1184
9066
+ },
9067
+ {
9068
+ "epoch": 1.3524338963723184,
9069
+ "grad_norm": 51.29532599637972,
9070
+ "learning_rate": 5.254616266367591e-08,
9071
+ "logits/chosen": -1.2815008163452148,
9072
+ "logits/rejected": -1.3232269287109375,
9073
+ "logps/chosen": -141.40736389160156,
9074
+ "logps/rejected": -170.84877014160156,
9075
+ "loss": 0.4018,
9076
+ "rewards/accuracies": 0.78125,
9077
+ "rewards/chosen": -0.3718317449092865,
9078
+ "rewards/margins": 0.9131482839584351,
9079
+ "rewards/rejected": -1.284980058670044,
9080
+ "step": 1186
9081
+ },
9082
+ {
9083
+ "epoch": 1.3547145606157793,
9084
+ "grad_norm": 68.71395143271354,
9085
+ "learning_rate": 5.2211719864976855e-08,
9086
+ "logits/chosen": -1.3111282587051392,
9087
+ "logits/rejected": -1.2713509798049927,
9088
+ "logps/chosen": -168.83644104003906,
9089
+ "logps/rejected": -189.7281951904297,
9090
+ "loss": 0.4716,
9091
+ "rewards/accuracies": 0.71875,
9092
+ "rewards/chosen": -0.8040088415145874,
9093
+ "rewards/margins": 0.7993437647819519,
9094
+ "rewards/rejected": -1.603352665901184,
9095
+ "step": 1188
9096
+ },
9097
+ {
9098
+ "epoch": 1.3569952248592403,
9099
+ "grad_norm": 47.5011022235632,
9100
+ "learning_rate": 5.1877968356681365e-08,
9101
+ "logits/chosen": -1.3059942722320557,
9102
+ "logits/rejected": -1.3891113996505737,
9103
+ "logps/chosen": -139.248046875,
9104
+ "logps/rejected": -177.55447387695312,
9105
+ "loss": 0.4136,
9106
+ "rewards/accuracies": 0.96875,
9107
+ "rewards/chosen": -0.3202630281448364,
9108
+ "rewards/margins": 1.2098579406738281,
9109
+ "rewards/rejected": -1.530120849609375,
9110
+ "step": 1190
9111
+ },
9112
+ {
9113
+ "epoch": 1.359275889102701,
9114
+ "grad_norm": 61.28282316729743,
9115
+ "learning_rate": 5.154491296673499e-08,
9116
+ "logits/chosen": -1.3505187034606934,
9117
+ "logits/rejected": -1.3525949716567993,
9118
+ "logps/chosen": -151.95126342773438,
9119
+ "logps/rejected": -170.53799438476562,
9120
+ "loss": 0.4111,
9121
+ "rewards/accuracies": 0.78125,
9122
+ "rewards/chosen": -0.29294249415397644,
9123
+ "rewards/margins": 0.8566387295722961,
9124
+ "rewards/rejected": -1.1495811939239502,
9125
+ "step": 1192
9126
+ },
9127
+ {
9128
+ "epoch": 1.361556553346162,
9129
+ "grad_norm": 50.10110632099787,
9130
+ "learning_rate": 5.121255851301345e-08,
9131
+ "logits/chosen": -1.182908058166504,
9132
+ "logits/rejected": -1.268027901649475,
9133
+ "logps/chosen": -184.67160034179688,
9134
+ "logps/rejected": -275.9751892089844,
9135
+ "loss": 0.4072,
9136
+ "rewards/accuracies": 0.96875,
9137
+ "rewards/chosen": -0.4911373257637024,
9138
+ "rewards/margins": 1.55191171169281,
9139
+ "rewards/rejected": -2.043048858642578,
9140
+ "step": 1194
9141
+ },
9142
+ {
9143
+ "epoch": 1.363837217589623,
9144
+ "grad_norm": 54.80752104995682,
9145
+ "learning_rate": 5.088090980325297e-08,
9146
+ "logits/chosen": -1.2852065563201904,
9147
+ "logits/rejected": -1.3752135038375854,
9148
+ "logps/chosen": -202.42112731933594,
9149
+ "logps/rejected": -243.62234497070312,
9150
+ "loss": 0.4245,
9151
+ "rewards/accuracies": 0.875,
9152
+ "rewards/chosen": -0.540120542049408,
9153
+ "rewards/margins": 1.2409517765045166,
9154
+ "rewards/rejected": -1.7810722589492798,
9155
+ "step": 1196
9156
+ },
9157
+ {
9158
+ "epoch": 1.366117881833084,
9159
+ "grad_norm": 57.983612303856106,
9160
+ "learning_rate": 5.054997163498065e-08,
9161
+ "logits/chosen": -1.4842596054077148,
9162
+ "logits/rejected": -1.454995036125183,
9163
+ "logps/chosen": -213.9556884765625,
9164
+ "logps/rejected": -208.71621704101562,
9165
+ "loss": 0.4446,
9166
+ "rewards/accuracies": 0.8125,
9167
+ "rewards/chosen": -0.512903094291687,
9168
+ "rewards/margins": 0.7354660034179688,
9169
+ "rewards/rejected": -1.2483689785003662,
9170
+ "step": 1198
9171
+ },
9172
+ {
9173
+ "epoch": 1.3683985460765449,
9174
+ "grad_norm": 56.17410591987438,
9175
+ "learning_rate": 5.0219748795445214e-08,
9176
+ "logits/chosen": -1.39608895778656,
9177
+ "logits/rejected": -1.4197230339050293,
9178
+ "logps/chosen": -151.9307403564453,
9179
+ "logps/rejected": -177.78945922851562,
9180
+ "loss": 0.3957,
9181
+ "rewards/accuracies": 0.90625,
9182
+ "rewards/chosen": -0.07908481359481812,
9183
+ "rewards/margins": 1.1620317697525024,
9184
+ "rewards/rejected": -1.2411167621612549,
9185
+ "step": 1200
9186
+ },
9187
+ {
9188
+ "epoch": 1.3683985460765449,
9189
+ "eval_logits/chosen": -1.3583488464355469,
9190
+ "eval_logits/rejected": -1.341153860092163,
9191
+ "eval_logps/chosen": -132.06036376953125,
9192
+ "eval_logps/rejected": -138.48941040039062,
9193
+ "eval_loss": 0.5543201565742493,
9194
+ "eval_rewards/accuracies": 0.7599999904632568,
9195
+ "eval_rewards/chosen": -0.2162403166294098,
9196
+ "eval_rewards/margins": 0.42642509937286377,
9197
+ "eval_rewards/rejected": -0.6426653861999512,
9198
+ "eval_runtime": 20.2372,
9199
+ "eval_samples_per_second": 4.941,
9200
+ "eval_steps_per_second": 1.235,
9201
+ "step": 1200
9202
  }
9203
  ],
9204
  "logging_steps": 2,