RyanYr commited on
Commit
0de77b2
1 Parent(s): ab1e1e3

Training in progress, step 900, checkpoint

Browse files
last-checkpoint/global_step900/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 sha256:0c113394da562d0875b8009f51b68aae623a91a5577187af4ba32ba873468692
3
+ size 24090788996
last-checkpoint/global_step900/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 sha256:870b071d0d1e36cb8905143e9ad8ab74c53a9fae0c34e80ee00817425f7427fc
3
+ size 24090788996
last-checkpoint/global_step900/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 sha256:7b71e15eb85a6943a678c1de19ef54bc9c1bf655f406af0fea830b9f6aa12450
3
+ size 24090788996
last-checkpoint/global_step900/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 sha256:400a877d1b504e13fbf6608fee0ea36dfbb49781aeed37eb3e3873046b4e4889
3
+ size 24090788996
last-checkpoint/global_step900/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 sha256:03aa56ecb1f68ecc90420d814e9b3c1abe084d46234c983e63e13cea393721fe
3
+ size 150693
last-checkpoint/global_step900/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:c1a7593ce7a3927186a34d438d43d481f0d1304ddf82e1e47e0bce89d453d63f
3
+ size 150693
last-checkpoint/global_step900/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:cab96a0f726e6ca8e6b7623eba72be6992c92e0456aaee53650a2b9e5645449b
3
+ size 150693
last-checkpoint/global_step900/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:e0d4cf9da5f72c8d9e5559c8487fcf01752eaa93e0b3be1fcdfbc773e767acc5
3
+ size 150693
last-checkpoint/latest CHANGED
@@ -1 +1 @@
1
- global_step600
 
1
+ global_step900
last-checkpoint/model-00001-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:48948a0d9104661249099a4014490ab23bd0ad9a31f6471dee9f9f094a3d63bf
3
  size 4976698672
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b0a49a34d9ad1fd47bdf6e82c652236ea730bca3941b775b4e3c7e709fa16644
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 sha256:8b24fa74f6060bb886c4ad908a9e1cbc5b1d32700a032f508c04604951a8181f
3
  size 4999802720
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b143a16d86f8b74d3600f817bcab8d67803ee67ee704a772e2f80da18faaed43
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 sha256:4f1fb4fda3abf0c7433b1013f78feb462c78a1d36b3a3336e8e2ed0f8c341706
3
  size 4915916176
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d11525e7ca69eeb3664967f7a14fed39b116e52ed11763c07abab32302315d63
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 sha256:1a42d0ba56615ac5b6ca669d046a3ac05a8adeee3b01ebef26cde18b002e96c8
3
  size 1168138808
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:43b2eb34fb6b2691e0523457cc256edd2fd64a363125e2d3763c744933525e52
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:ef002048764051a71fb00f8f978e9ec32b780dc850bdb059af362cc56494234b
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bdd1f02cb20d3f4f7e0dd26fea62af57e5e71316163f926a28ed6cf89a9f3777
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 sha256:37194a6d48612e1a46a2d5d317ead97c70d9fc4569b0118fcd5f84c3dc9daa5a
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc6d54ba2aa85e2f895439a1b787ec947b848a1c34ea5a3a28821572bf2b9fec
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 sha256:17c179483659a784aa1ace2427daff48c556a6bcc3c330e6f3274e4dc95e4b49
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8b6927d26551cddd8e35b34b43e79bd58f8b6027b6a481bb6a563a3652addeb4
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:b56857c9b117629f35af2c3d64f522d33a9d8aa94faa81ec6956380a895118c4
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f8295b1be8e66b4b30cb905dc48cfc717c027e427937b8142d00ae9de8106c6a
3
  size 15024
last-checkpoint/scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8d8f9b18fc83c21830420fd2e6d55afd183068e9c7f7ec7447233ce473235b6a
3
  size 1064
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5ddbeb3cc81a6801e3a5f051d8eb975a0ae4e8af69a917df5cb66bf05d97ea71
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": 0.6948968512486428,
5
  "eval_steps": 100,
6
- "global_step": 600,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
@@ -4603,6 +4603,2304 @@
4603
  "eval_samples_per_second": 4.01,
4604
  "eval_steps_per_second": 1.002,
4605
  "step": 600
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4606
  }
4607
  ],
4608
  "logging_steps": 2,
 
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
+ "epoch": 1.0423452768729642,
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,
 
4603
  "eval_samples_per_second": 4.01,
4604
  "eval_steps_per_second": 1.002,
4605
  "step": 600
4606
+ },
4607
+ {
4608
+ "epoch": 0.6972131740861383,
4609
+ "grad_norm": 65.21565931204394,
4610
+ "learning_rate": 1.5653707999357954e-07,
4611
+ "logits/chosen": -1.246476411819458,
4612
+ "logits/rejected": -1.3448492288589478,
4613
+ "logps/chosen": -145.94290161132812,
4614
+ "logps/rejected": -167.75537109375,
4615
+ "loss": 0.5257,
4616
+ "rewards/accuracies": 0.71875,
4617
+ "rewards/chosen": 0.09868886321783066,
4618
+ "rewards/margins": 0.4459027051925659,
4619
+ "rewards/rejected": -0.3472138047218323,
4620
+ "step": 602
4621
+ },
4622
+ {
4623
+ "epoch": 0.6995294969236338,
4624
+ "grad_norm": 83.43610856907603,
4625
+ "learning_rate": 1.5621792541582965e-07,
4626
+ "logits/chosen": -1.3476929664611816,
4627
+ "logits/rejected": -1.3578147888183594,
4628
+ "logps/chosen": -147.28909301757812,
4629
+ "logps/rejected": -151.40380859375,
4630
+ "loss": 0.6078,
4631
+ "rewards/accuracies": 0.6875,
4632
+ "rewards/chosen": -0.13675910234451294,
4633
+ "rewards/margins": 0.5534166097640991,
4634
+ "rewards/rejected": -0.6901756525039673,
4635
+ "step": 604
4636
+ },
4637
+ {
4638
+ "epoch": 0.7018458197611293,
4639
+ "grad_norm": 83.4114443803894,
4640
+ "learning_rate": 1.558979313917613e-07,
4641
+ "logits/chosen": -1.3711118698120117,
4642
+ "logits/rejected": -1.4022268056869507,
4643
+ "logps/chosen": -147.4836883544922,
4644
+ "logps/rejected": -200.58444213867188,
4645
+ "loss": 0.5579,
4646
+ "rewards/accuracies": 0.75,
4647
+ "rewards/chosen": -0.20421257615089417,
4648
+ "rewards/margins": 1.2417640686035156,
4649
+ "rewards/rejected": -1.4459766149520874,
4650
+ "step": 606
4651
+ },
4652
+ {
4653
+ "epoch": 0.7041621425986246,
4654
+ "grad_norm": 86.87494281849429,
4655
+ "learning_rate": 1.5557710269952668e-07,
4656
+ "logits/chosen": -1.2246983051300049,
4657
+ "logits/rejected": -1.257164716720581,
4658
+ "logps/chosen": -126.96261596679688,
4659
+ "logps/rejected": -146.96583557128906,
4660
+ "loss": 0.618,
4661
+ "rewards/accuracies": 0.5625,
4662
+ "rewards/chosen": -0.23703543841838837,
4663
+ "rewards/margins": 0.6981402635574341,
4664
+ "rewards/rejected": -0.9351757168769836,
4665
+ "step": 608
4666
+ },
4667
+ {
4668
+ "epoch": 0.7064784654361201,
4669
+ "grad_norm": 78.38055223849892,
4670
+ "learning_rate": 1.552554441297413e-07,
4671
+ "logits/chosen": -1.3179811239242554,
4672
+ "logits/rejected": -1.3713396787643433,
4673
+ "logps/chosen": -152.94192504882812,
4674
+ "logps/rejected": -176.10769653320312,
4675
+ "loss": 0.6449,
4676
+ "rewards/accuracies": 0.6875,
4677
+ "rewards/chosen": -0.025582734495401382,
4678
+ "rewards/margins": 0.6770236492156982,
4679
+ "rewards/rejected": -0.7026063799858093,
4680
+ "step": 610
4681
+ },
4682
+ {
4683
+ "epoch": 0.7087947882736156,
4684
+ "grad_norm": 74.133399784032,
4685
+ "learning_rate": 1.549329604854124e-07,
4686
+ "logits/chosen": -1.3148674964904785,
4687
+ "logits/rejected": -1.327743649482727,
4688
+ "logps/chosen": -139.4869384765625,
4689
+ "logps/rejected": -159.59814453125,
4690
+ "loss": 0.5677,
4691
+ "rewards/accuracies": 0.75,
4692
+ "rewards/chosen": -0.14542880654335022,
4693
+ "rewards/margins": 0.8610043525695801,
4694
+ "rewards/rejected": -1.0064332485198975,
4695
+ "step": 612
4696
+ },
4697
+ {
4698
+ "epoch": 0.7111111111111111,
4699
+ "grad_norm": 112.65020221864026,
4700
+ "learning_rate": 1.5460965658186714e-07,
4701
+ "logits/chosen": -1.3466724157333374,
4702
+ "logits/rejected": -1.3887736797332764,
4703
+ "logps/chosen": -151.4530029296875,
4704
+ "logps/rejected": -177.39657592773438,
4705
+ "loss": 0.6323,
4706
+ "rewards/accuracies": 0.78125,
4707
+ "rewards/chosen": -0.09504815936088562,
4708
+ "rewards/margins": 0.42471885681152344,
4709
+ "rewards/rejected": -0.5197670459747314,
4710
+ "step": 614
4711
+ },
4712
+ {
4713
+ "epoch": 0.7134274339486066,
4714
+ "grad_norm": 76.25616510726566,
4715
+ "learning_rate": 1.54285537246681e-07,
4716
+ "logits/chosen": -1.24288010597229,
4717
+ "logits/rejected": -1.2770837545394897,
4718
+ "logps/chosen": -111.17274475097656,
4719
+ "logps/rejected": -145.47189331054688,
4720
+ "loss": 0.5485,
4721
+ "rewards/accuracies": 0.625,
4722
+ "rewards/chosen": 0.026386726647615433,
4723
+ "rewards/margins": 0.590622067451477,
4724
+ "rewards/rejected": -0.5642353296279907,
4725
+ "step": 616
4726
+ },
4727
+ {
4728
+ "epoch": 0.715743756786102,
4729
+ "grad_norm": 138.34950058984242,
4730
+ "learning_rate": 1.539606073196053e-07,
4731
+ "logits/chosen": -1.3290059566497803,
4732
+ "logits/rejected": -1.3147568702697754,
4733
+ "logps/chosen": -189.2161865234375,
4734
+ "logps/rejected": -208.13873291015625,
4735
+ "loss": 0.6596,
4736
+ "rewards/accuracies": 0.625,
4737
+ "rewards/chosen": -0.6345406174659729,
4738
+ "rewards/margins": 0.8859498500823975,
4739
+ "rewards/rejected": -1.5204904079437256,
4740
+ "step": 618
4741
+ },
4742
+ {
4743
+ "epoch": 0.7180600796235975,
4744
+ "grad_norm": 150.2269252956893,
4745
+ "learning_rate": 1.536348716524952e-07,
4746
+ "logits/chosen": -1.2324703931808472,
4747
+ "logits/rejected": -1.3258979320526123,
4748
+ "logps/chosen": -115.2970199584961,
4749
+ "logps/rejected": -129.09571838378906,
4750
+ "loss": 0.6866,
4751
+ "rewards/accuracies": 0.59375,
4752
+ "rewards/chosen": 0.001235203817486763,
4753
+ "rewards/margins": 0.2507448196411133,
4754
+ "rewards/rejected": -0.24950963258743286,
4755
+ "step": 620
4756
+ },
4757
+ {
4758
+ "epoch": 0.720376402461093,
4759
+ "grad_norm": 87.5580010456215,
4760
+ "learning_rate": 1.5330833510923716e-07,
4761
+ "logits/chosen": -1.257300615310669,
4762
+ "logits/rejected": -1.2330734729766846,
4763
+ "logps/chosen": -146.174072265625,
4764
+ "logps/rejected": -167.9958038330078,
4765
+ "loss": 0.5343,
4766
+ "rewards/accuracies": 0.78125,
4767
+ "rewards/chosen": -0.09243424236774445,
4768
+ "rewards/margins": 0.7605526447296143,
4769
+ "rewards/rejected": -0.8529868125915527,
4770
+ "step": 622
4771
+ },
4772
+ {
4773
+ "epoch": 0.7226927252985885,
4774
+ "grad_norm": 80.10448855296569,
4775
+ "learning_rate": 1.529810025656764e-07,
4776
+ "logits/chosen": -1.3765792846679688,
4777
+ "logits/rejected": -1.3748364448547363,
4778
+ "logps/chosen": -108.47747802734375,
4779
+ "logps/rejected": -115.16337585449219,
4780
+ "loss": 0.5844,
4781
+ "rewards/accuracies": 0.625,
4782
+ "rewards/chosen": 0.10373257100582123,
4783
+ "rewards/margins": 0.365066796541214,
4784
+ "rewards/rejected": -0.26133421063423157,
4785
+ "step": 624
4786
+ },
4787
+ {
4788
+ "epoch": 0.7250090481360839,
4789
+ "grad_norm": 83.63175479164269,
4790
+ "learning_rate": 1.5265287890954386e-07,
4791
+ "logits/chosen": -1.2026071548461914,
4792
+ "logits/rejected": -1.2504820823669434,
4793
+ "logps/chosen": -132.87338256835938,
4794
+ "logps/rejected": -209.71510314941406,
4795
+ "loss": 0.5652,
4796
+ "rewards/accuracies": 0.71875,
4797
+ "rewards/chosen": -0.26427173614501953,
4798
+ "rewards/margins": 0.8983727693557739,
4799
+ "rewards/rejected": -1.1626445055007935,
4800
+ "step": 626
4801
+ },
4802
+ {
4803
+ "epoch": 0.7273253709735794,
4804
+ "grad_norm": 99.26984323238828,
4805
+ "learning_rate": 1.523239690403835e-07,
4806
+ "logits/chosen": -1.2745357751846313,
4807
+ "logits/rejected": -1.3479351997375488,
4808
+ "logps/chosen": -141.58119201660156,
4809
+ "logps/rejected": -179.41603088378906,
4810
+ "loss": 0.5868,
4811
+ "rewards/accuracies": 0.5625,
4812
+ "rewards/chosen": -0.12304151058197021,
4813
+ "rewards/margins": 0.5740844011306763,
4814
+ "rewards/rejected": -0.6971259117126465,
4815
+ "step": 628
4816
+ },
4817
+ {
4818
+ "epoch": 0.7296416938110749,
4819
+ "grad_norm": 68.60907419248075,
4820
+ "learning_rate": 1.5199427786947898e-07,
4821
+ "logits/chosen": -1.2993189096450806,
4822
+ "logits/rejected": -1.268160104751587,
4823
+ "logps/chosen": -150.71392822265625,
4824
+ "logps/rejected": -175.40994262695312,
4825
+ "loss": 0.5011,
4826
+ "rewards/accuracies": 0.875,
4827
+ "rewards/chosen": -0.05146133154630661,
4828
+ "rewards/margins": 0.942064106464386,
4829
+ "rewards/rejected": -0.993525505065918,
4830
+ "step": 630
4831
+ },
4832
+ {
4833
+ "epoch": 0.7319580166485704,
4834
+ "grad_norm": 103.51182599026306,
4835
+ "learning_rate": 1.5166381031978043e-07,
4836
+ "logits/chosen": -1.2612452507019043,
4837
+ "logits/rejected": -1.313174843788147,
4838
+ "logps/chosen": -160.19854736328125,
4839
+ "logps/rejected": -210.183349609375,
4840
+ "loss": 0.5902,
4841
+ "rewards/accuracies": 0.6875,
4842
+ "rewards/chosen": -0.494692325592041,
4843
+ "rewards/margins": 1.7965799570083618,
4844
+ "rewards/rejected": -2.2912724018096924,
4845
+ "step": 632
4846
+ },
4847
+ {
4848
+ "epoch": 0.7342743394860659,
4849
+ "grad_norm": 62.22603011415776,
4850
+ "learning_rate": 1.5133257132583073e-07,
4851
+ "logits/chosen": -1.26754629611969,
4852
+ "logits/rejected": -1.2881284952163696,
4853
+ "logps/chosen": -159.5138702392578,
4854
+ "logps/rejected": -161.47116088867188,
4855
+ "loss": 0.5607,
4856
+ "rewards/accuracies": 0.75,
4857
+ "rewards/chosen": -0.3782161772251129,
4858
+ "rewards/margins": 0.5586279630661011,
4859
+ "rewards/rejected": -0.9368441700935364,
4860
+ "step": 634
4861
+ },
4862
+ {
4863
+ "epoch": 0.7365906623235613,
4864
+ "grad_norm": 96.15037328481357,
4865
+ "learning_rate": 1.5100056583369205e-07,
4866
+ "logits/chosen": -1.2539037466049194,
4867
+ "logits/rejected": -1.3549363613128662,
4868
+ "logps/chosen": -155.125732421875,
4869
+ "logps/rejected": -164.1871795654297,
4870
+ "loss": 0.5928,
4871
+ "rewards/accuracies": 0.65625,
4872
+ "rewards/chosen": -0.4689493477344513,
4873
+ "rewards/margins": 0.17761842906475067,
4874
+ "rewards/rejected": -0.6465678215026855,
4875
+ "step": 636
4876
+ },
4877
+ {
4878
+ "epoch": 0.7389069851610568,
4879
+ "grad_norm": 103.33944083718919,
4880
+ "learning_rate": 1.5066779880087195e-07,
4881
+ "logits/chosen": -1.2627050876617432,
4882
+ "logits/rejected": -1.29320228099823,
4883
+ "logps/chosen": -166.43197631835938,
4884
+ "logps/rejected": -198.9180450439453,
4885
+ "loss": 0.5905,
4886
+ "rewards/accuracies": 0.75,
4887
+ "rewards/chosen": 0.11970967054367065,
4888
+ "rewards/margins": 0.9318963289260864,
4889
+ "rewards/rejected": -0.8121867179870605,
4890
+ "step": 638
4891
+ },
4892
+ {
4893
+ "epoch": 0.7412233079985523,
4894
+ "grad_norm": 99.08292249787714,
4895
+ "learning_rate": 1.503342751962493e-07,
4896
+ "logits/chosen": -1.347541332244873,
4897
+ "logits/rejected": -1.396866798400879,
4898
+ "logps/chosen": -115.1908187866211,
4899
+ "logps/rejected": -140.8692169189453,
4900
+ "loss": 0.6687,
4901
+ "rewards/accuracies": 0.5625,
4902
+ "rewards/chosen": -0.448566734790802,
4903
+ "rewards/margins": 0.19808316230773926,
4904
+ "rewards/rejected": -0.6466498374938965,
4905
+ "step": 640
4906
+ },
4907
+ {
4908
+ "epoch": 0.7435396308360478,
4909
+ "grad_norm": 100.95301107774017,
4910
+ "learning_rate": 1.5e-07,
4911
+ "logits/chosen": -1.3385045528411865,
4912
+ "logits/rejected": -1.3928390741348267,
4913
+ "logps/chosen": -161.37307739257812,
4914
+ "logps/rejected": -180.67625427246094,
4915
+ "loss": 0.622,
4916
+ "rewards/accuracies": 0.5625,
4917
+ "rewards/chosen": -0.28616487979888916,
4918
+ "rewards/margins": 0.43838122487068176,
4919
+ "rewards/rejected": -0.7245461344718933,
4920
+ "step": 642
4921
+ },
4922
+ {
4923
+ "epoch": 0.7458559536735433,
4924
+ "grad_norm": 101.02506488024277,
4925
+ "learning_rate": 1.4966497820352286e-07,
4926
+ "logits/chosen": -1.2145590782165527,
4927
+ "logits/rejected": -1.2937660217285156,
4928
+ "logps/chosen": -158.820068359375,
4929
+ "logps/rejected": -178.55609130859375,
4930
+ "loss": 0.6347,
4931
+ "rewards/accuracies": 0.53125,
4932
+ "rewards/chosen": -0.7363303303718567,
4933
+ "rewards/margins": 0.2016220986843109,
4934
+ "rewards/rejected": -0.9379523992538452,
4935
+ "step": 644
4936
+ },
4937
+ {
4938
+ "epoch": 0.7481722765110387,
4939
+ "grad_norm": 87.43579693715355,
4940
+ "learning_rate": 1.493292148093649e-07,
4941
+ "logits/chosen": -1.3464206457138062,
4942
+ "logits/rejected": -1.3654189109802246,
4943
+ "logps/chosen": -151.60321044921875,
4944
+ "logps/rejected": -166.5888214111328,
4945
+ "loss": 0.5879,
4946
+ "rewards/accuracies": 0.6875,
4947
+ "rewards/chosen": -0.22216303646564484,
4948
+ "rewards/margins": 0.5710839629173279,
4949
+ "rewards/rejected": -0.7932470440864563,
4950
+ "step": 646
4951
+ },
4952
+ {
4953
+ "epoch": 0.7504885993485342,
4954
+ "grad_norm": 89.88093815940675,
4955
+ "learning_rate": 1.4899271483114656e-07,
4956
+ "logits/chosen": -1.2944061756134033,
4957
+ "logits/rejected": -1.2952131032943726,
4958
+ "logps/chosen": -87.6488037109375,
4959
+ "logps/rejected": -96.55010223388672,
4960
+ "loss": 0.8793,
4961
+ "rewards/accuracies": 0.65625,
4962
+ "rewards/chosen": -0.36783769726753235,
4963
+ "rewards/margins": 0.2082243710756302,
4964
+ "rewards/rejected": -0.576062023639679,
4965
+ "step": 648
4966
+ },
4967
+ {
4968
+ "epoch": 0.7528049221860297,
4969
+ "grad_norm": 68.16473110876538,
4970
+ "learning_rate": 1.4865548329348716e-07,
4971
+ "logits/chosen": -1.2365565299987793,
4972
+ "logits/rejected": -1.2194610834121704,
4973
+ "logps/chosen": -101.74496459960938,
4974
+ "logps/rejected": -115.47993469238281,
4975
+ "loss": 0.5926,
4976
+ "rewards/accuracies": 0.78125,
4977
+ "rewards/chosen": 0.09891890734434128,
4978
+ "rewards/margins": 0.535317063331604,
4979
+ "rewards/rejected": -0.43639814853668213,
4980
+ "step": 650
4981
+ },
4982
+ {
4983
+ "epoch": 0.7551212450235252,
4984
+ "grad_norm": 126.64604196312857,
4985
+ "learning_rate": 1.4831752523192947e-07,
4986
+ "logits/chosen": -1.3086532354354858,
4987
+ "logits/rejected": -1.347178339958191,
4988
+ "logps/chosen": -123.16075134277344,
4989
+ "logps/rejected": -140.86134338378906,
4990
+ "loss": 0.6408,
4991
+ "rewards/accuracies": 0.53125,
4992
+ "rewards/chosen": 0.06762253493070602,
4993
+ "rewards/margins": 0.4483943581581116,
4994
+ "rewards/rejected": -0.38077181577682495,
4995
+ "step": 652
4996
+ },
4997
+ {
4998
+ "epoch": 0.7574375678610207,
4999
+ "grad_norm": 87.13600344485694,
5000
+ "learning_rate": 1.4797884569286482e-07,
5001
+ "logits/chosen": -1.223707675933838,
5002
+ "logits/rejected": -1.3276126384735107,
5003
+ "logps/chosen": -154.83677673339844,
5004
+ "logps/rejected": -161.1993865966797,
5005
+ "loss": 0.561,
5006
+ "rewards/accuracies": 0.65625,
5007
+ "rewards/chosen": 0.12348895519971848,
5008
+ "rewards/margins": 0.5256034135818481,
5009
+ "rewards/rejected": -0.40211451053619385,
5010
+ "step": 654
5011
+ },
5012
+ {
5013
+ "epoch": 0.7597538906985161,
5014
+ "grad_norm": 77.97523563671065,
5015
+ "learning_rate": 1.476394497334577e-07,
5016
+ "logits/chosen": -1.2289892435073853,
5017
+ "logits/rejected": -1.2923333644866943,
5018
+ "logps/chosen": -133.74855041503906,
5019
+ "logps/rejected": -168.71890258789062,
5020
+ "loss": 0.5593,
5021
+ "rewards/accuracies": 0.625,
5022
+ "rewards/chosen": 0.054568979889154434,
5023
+ "rewards/margins": 1.1155064105987549,
5024
+ "rewards/rejected": -1.0609374046325684,
5025
+ "step": 656
5026
+ },
5027
+ {
5028
+ "epoch": 0.7620702135360116,
5029
+ "grad_norm": 86.95243414887648,
5030
+ "learning_rate": 1.4729934242157002e-07,
5031
+ "logits/chosen": -1.2043237686157227,
5032
+ "logits/rejected": -1.298114538192749,
5033
+ "logps/chosen": -147.36212158203125,
5034
+ "logps/rejected": -175.49537658691406,
5035
+ "loss": 0.5916,
5036
+ "rewards/accuracies": 0.84375,
5037
+ "rewards/chosen": 0.04841674864292145,
5038
+ "rewards/margins": 0.7525838613510132,
5039
+ "rewards/rejected": -0.7041671872138977,
5040
+ "step": 658
5041
+ },
5042
+ {
5043
+ "epoch": 0.7643865363735071,
5044
+ "grad_norm": 75.65437165493766,
5045
+ "learning_rate": 1.4695852883568577e-07,
5046
+ "logits/chosen": -1.296444058418274,
5047
+ "logits/rejected": -1.297696590423584,
5048
+ "logps/chosen": -139.0203399658203,
5049
+ "logps/rejected": -136.33407592773438,
5050
+ "loss": 0.664,
5051
+ "rewards/accuracies": 0.5625,
5052
+ "rewards/chosen": -0.03904925286769867,
5053
+ "rewards/margins": 0.23676946759223938,
5054
+ "rewards/rejected": -0.27581873536109924,
5055
+ "step": 660
5056
+ },
5057
+ {
5058
+ "epoch": 0.7667028592110026,
5059
+ "grad_norm": 74.56863133576563,
5060
+ "learning_rate": 1.46617014064835e-07,
5061
+ "logits/chosen": -1.31462562084198,
5062
+ "logits/rejected": -1.3569142818450928,
5063
+ "logps/chosen": -84.67462158203125,
5064
+ "logps/rejected": -85.39293670654297,
5065
+ "loss": 0.6178,
5066
+ "rewards/accuracies": 0.71875,
5067
+ "rewards/chosen": 0.08590921759605408,
5068
+ "rewards/margins": 0.2730802893638611,
5069
+ "rewards/rejected": -0.1871711015701294,
5070
+ "step": 662
5071
+ },
5072
+ {
5073
+ "epoch": 0.7690191820484981,
5074
+ "grad_norm": 65.1421142926074,
5075
+ "learning_rate": 1.4627480320851773e-07,
5076
+ "logits/chosen": -1.2603397369384766,
5077
+ "logits/rejected": -1.3087056875228882,
5078
+ "logps/chosen": -94.25755310058594,
5079
+ "logps/rejected": -124.65604400634766,
5080
+ "loss": 0.5517,
5081
+ "rewards/accuracies": 0.5625,
5082
+ "rewards/chosen": 0.13362671434879303,
5083
+ "rewards/margins": 0.27784639596939087,
5084
+ "rewards/rejected": -0.14421969652175903,
5085
+ "step": 664
5086
+ },
5087
+ {
5088
+ "epoch": 0.7713355048859935,
5089
+ "grad_norm": 71.68830752075247,
5090
+ "learning_rate": 1.459319013766281e-07,
5091
+ "logits/chosen": -1.2157174348831177,
5092
+ "logits/rejected": -1.2898728847503662,
5093
+ "logps/chosen": -115.97237396240234,
5094
+ "logps/rejected": -161.3004913330078,
5095
+ "loss": 0.5697,
5096
+ "rewards/accuracies": 0.5625,
5097
+ "rewards/chosen": 0.03516310453414917,
5098
+ "rewards/margins": 0.6528601050376892,
5099
+ "rewards/rejected": -0.6176969408988953,
5100
+ "step": 666
5101
+ },
5102
+ {
5103
+ "epoch": 0.773651827723489,
5104
+ "grad_norm": 79.36186918181832,
5105
+ "learning_rate": 1.4558831368937782e-07,
5106
+ "logits/chosen": -1.4229352474212646,
5107
+ "logits/rejected": -1.4072250127792358,
5108
+ "logps/chosen": -177.0735626220703,
5109
+ "logps/rejected": -197.98715209960938,
5110
+ "loss": 0.5665,
5111
+ "rewards/accuracies": 0.78125,
5112
+ "rewards/chosen": 0.10195475816726685,
5113
+ "rewards/margins": 0.687512993812561,
5114
+ "rewards/rejected": -0.5855582356452942,
5115
+ "step": 668
5116
+ },
5117
+ {
5118
+ "epoch": 0.7759681505609844,
5119
+ "grad_norm": 87.8388965497945,
5120
+ "learning_rate": 1.4524404527721975e-07,
5121
+ "logits/chosen": -1.3583612442016602,
5122
+ "logits/rejected": -1.389435052871704,
5123
+ "logps/chosen": -122.91793823242188,
5124
+ "logps/rejected": -143.0467071533203,
5125
+ "loss": 0.5898,
5126
+ "rewards/accuracies": 0.65625,
5127
+ "rewards/chosen": 0.04354669526219368,
5128
+ "rewards/margins": 0.41671624779701233,
5129
+ "rewards/rejected": -0.37316957116127014,
5130
+ "step": 670
5131
+ },
5132
+ {
5133
+ "epoch": 0.77828447339848,
5134
+ "grad_norm": 81.7257176752688,
5135
+ "learning_rate": 1.4489910128077144e-07,
5136
+ "logits/chosen": -1.247169017791748,
5137
+ "logits/rejected": -1.233970046043396,
5138
+ "logps/chosen": -160.17123413085938,
5139
+ "logps/rejected": -161.0812225341797,
5140
+ "loss": 0.5867,
5141
+ "rewards/accuracies": 0.625,
5142
+ "rewards/chosen": 0.1097157821059227,
5143
+ "rewards/margins": 0.3292488753795624,
5144
+ "rewards/rejected": -0.21953308582305908,
5145
+ "step": 672
5146
+ },
5147
+ {
5148
+ "epoch": 0.7806007962359753,
5149
+ "grad_norm": 82.58162746498225,
5150
+ "learning_rate": 1.4455348685073824e-07,
5151
+ "logits/chosen": -1.3045804500579834,
5152
+ "logits/rejected": -1.3406845331192017,
5153
+ "logps/chosen": -144.19696044921875,
5154
+ "logps/rejected": -166.00563049316406,
5155
+ "loss": 0.5414,
5156
+ "rewards/accuracies": 0.8125,
5157
+ "rewards/chosen": 0.17315542697906494,
5158
+ "rewards/margins": 0.8811919689178467,
5159
+ "rewards/rejected": -0.7080365419387817,
5160
+ "step": 674
5161
+ },
5162
+ {
5163
+ "epoch": 0.7829171190734708,
5164
+ "grad_norm": 70.90800914750002,
5165
+ "learning_rate": 1.4420720714783634e-07,
5166
+ "logits/chosen": -1.3308024406433105,
5167
+ "logits/rejected": -1.3253967761993408,
5168
+ "logps/chosen": -168.21568298339844,
5169
+ "logps/rejected": -170.13629150390625,
5170
+ "loss": 0.6002,
5171
+ "rewards/accuracies": 0.625,
5172
+ "rewards/chosen": 0.01426420547068119,
5173
+ "rewards/margins": 0.2677161991596222,
5174
+ "rewards/rejected": -0.25345203280448914,
5175
+ "step": 676
5176
+ },
5177
+ {
5178
+ "epoch": 0.7852334419109663,
5179
+ "grad_norm": 108.18465138210843,
5180
+ "learning_rate": 1.438602673427158e-07,
5181
+ "logits/chosen": -1.243054747581482,
5182
+ "logits/rejected": -1.284456729888916,
5183
+ "logps/chosen": -141.8882598876953,
5184
+ "logps/rejected": -142.9855499267578,
5185
+ "loss": 0.6341,
5186
+ "rewards/accuracies": 0.71875,
5187
+ "rewards/chosen": -0.468353807926178,
5188
+ "rewards/margins": 0.15223075449466705,
5189
+ "rewards/rejected": -0.6205846667289734,
5190
+ "step": 678
5191
+ },
5192
+ {
5193
+ "epoch": 0.7875497647484618,
5194
+ "grad_norm": 136.97239295907355,
5195
+ "learning_rate": 1.435126726158835e-07,
5196
+ "logits/chosen": -1.1808712482452393,
5197
+ "logits/rejected": -1.190900444984436,
5198
+ "logps/chosen": -108.58960723876953,
5199
+ "logps/rejected": -115.77843475341797,
5200
+ "loss": 0.6642,
5201
+ "rewards/accuracies": 0.5,
5202
+ "rewards/chosen": -0.1600310057401657,
5203
+ "rewards/margins": 0.3264339566230774,
5204
+ "rewards/rejected": -0.4864649176597595,
5205
+ "step": 680
5206
+ },
5207
+ {
5208
+ "epoch": 0.7898660875859573,
5209
+ "grad_norm": 77.26216998107256,
5210
+ "learning_rate": 1.431644281576254e-07,
5211
+ "logits/chosen": -1.352297067642212,
5212
+ "logits/rejected": -1.3604989051818848,
5213
+ "logps/chosen": -158.22714233398438,
5214
+ "logps/rejected": -179.4667510986328,
5215
+ "loss": 0.5683,
5216
+ "rewards/accuracies": 0.78125,
5217
+ "rewards/chosen": 0.03663470223546028,
5218
+ "rewards/margins": 0.6152007579803467,
5219
+ "rewards/rejected": -0.5785660147666931,
5220
+ "step": 682
5221
+ },
5222
+ {
5223
+ "epoch": 0.7921824104234527,
5224
+ "grad_norm": 86.99102850306988,
5225
+ "learning_rate": 1.4281553916792933e-07,
5226
+ "logits/chosen": -1.3430233001708984,
5227
+ "logits/rejected": -1.3200087547302246,
5228
+ "logps/chosen": -177.13780212402344,
5229
+ "logps/rejected": -197.31582641601562,
5230
+ "loss": 0.5711,
5231
+ "rewards/accuracies": 0.65625,
5232
+ "rewards/chosen": 0.08363021910190582,
5233
+ "rewards/margins": 0.7721484899520874,
5234
+ "rewards/rejected": -0.688518226146698,
5235
+ "step": 684
5236
+ },
5237
+ {
5238
+ "epoch": 0.7944987332609482,
5239
+ "grad_norm": 95.74498787413943,
5240
+ "learning_rate": 1.4246601085640734e-07,
5241
+ "logits/chosen": -1.3332422971725464,
5242
+ "logits/rejected": -1.3727816343307495,
5243
+ "logps/chosen": -146.84754943847656,
5244
+ "logps/rejected": -159.5667724609375,
5245
+ "loss": 0.5676,
5246
+ "rewards/accuracies": 0.71875,
5247
+ "rewards/chosen": -0.121219202876091,
5248
+ "rewards/margins": 0.5295783877372742,
5249
+ "rewards/rejected": -0.6507976055145264,
5250
+ "step": 686
5251
+ },
5252
+ {
5253
+ "epoch": 0.7968150560984437,
5254
+ "grad_norm": 77.80530486828445,
5255
+ "learning_rate": 1.421158484422177e-07,
5256
+ "logits/chosen": -1.1991747617721558,
5257
+ "logits/rejected": -1.193981409072876,
5258
+ "logps/chosen": -111.72264099121094,
5259
+ "logps/rejected": -151.81735229492188,
5260
+ "loss": 0.6152,
5261
+ "rewards/accuracies": 0.65625,
5262
+ "rewards/chosen": -0.2821239233016968,
5263
+ "rewards/margins": 0.8215171098709106,
5264
+ "rewards/rejected": -1.1036410331726074,
5265
+ "step": 688
5266
+ },
5267
+ {
5268
+ "epoch": 0.7991313789359392,
5269
+ "grad_norm": 78.69284003499466,
5270
+ "learning_rate": 1.417650571539872e-07,
5271
+ "logits/chosen": -1.184325933456421,
5272
+ "logits/rejected": -1.220619559288025,
5273
+ "logps/chosen": -117.9516372680664,
5274
+ "logps/rejected": -150.95596313476562,
5275
+ "loss": 0.6134,
5276
+ "rewards/accuracies": 0.5625,
5277
+ "rewards/chosen": -0.1240064799785614,
5278
+ "rewards/margins": 0.604076087474823,
5279
+ "rewards/rejected": -0.7280825972557068,
5280
+ "step": 690
5281
+ },
5282
+ {
5283
+ "epoch": 0.8014477017734347,
5284
+ "grad_norm": 101.94725129563409,
5285
+ "learning_rate": 1.4141364222973293e-07,
5286
+ "logits/chosen": -1.2512197494506836,
5287
+ "logits/rejected": -1.2510242462158203,
5288
+ "logps/chosen": -105.66606140136719,
5289
+ "logps/rejected": -126.79011535644531,
5290
+ "loss": 0.6816,
5291
+ "rewards/accuracies": 0.6875,
5292
+ "rewards/chosen": 0.07451184839010239,
5293
+ "rewards/margins": 0.3946147859096527,
5294
+ "rewards/rejected": -0.3201029300689697,
5295
+ "step": 692
5296
+ },
5297
+ {
5298
+ "epoch": 0.8037640246109301,
5299
+ "grad_norm": 92.22432532521196,
5300
+ "learning_rate": 1.410616089167842e-07,
5301
+ "logits/chosen": -1.3338427543640137,
5302
+ "logits/rejected": -1.3960174322128296,
5303
+ "logps/chosen": -151.6918487548828,
5304
+ "logps/rejected": -172.7644805908203,
5305
+ "loss": 0.6545,
5306
+ "rewards/accuracies": 0.78125,
5307
+ "rewards/chosen": 0.015777569264173508,
5308
+ "rewards/margins": 0.5224604606628418,
5309
+ "rewards/rejected": -0.5066828727722168,
5310
+ "step": 694
5311
+ },
5312
+ {
5313
+ "epoch": 0.8060803474484256,
5314
+ "grad_norm": 78.94912552978236,
5315
+ "learning_rate": 1.40708962471704e-07,
5316
+ "logits/chosen": -1.3187270164489746,
5317
+ "logits/rejected": -1.273497223854065,
5318
+ "logps/chosen": -149.56173706054688,
5319
+ "logps/rejected": -159.30621337890625,
5320
+ "loss": 0.5606,
5321
+ "rewards/accuracies": 0.71875,
5322
+ "rewards/chosen": -0.10468573868274689,
5323
+ "rewards/margins": 0.5250222682952881,
5324
+ "rewards/rejected": -0.6297080516815186,
5325
+ "step": 696
5326
+ },
5327
+ {
5328
+ "epoch": 0.8083966702859211,
5329
+ "grad_norm": 73.36588542547142,
5330
+ "learning_rate": 1.4035570816021066e-07,
5331
+ "logits/chosen": -1.1806639432907104,
5332
+ "logits/rejected": -1.2237762212753296,
5333
+ "logps/chosen": -124.83197784423828,
5334
+ "logps/rejected": -137.32818603515625,
5335
+ "loss": 0.5717,
5336
+ "rewards/accuracies": 0.75,
5337
+ "rewards/chosen": 0.13686442375183105,
5338
+ "rewards/margins": 0.6119322180747986,
5339
+ "rewards/rejected": -0.47506779432296753,
5340
+ "step": 698
5341
+ },
5342
+ {
5343
+ "epoch": 0.8107129931234166,
5344
+ "grad_norm": 90.26959143913993,
5345
+ "learning_rate": 1.4000185125709917e-07,
5346
+ "logits/chosen": -1.2998504638671875,
5347
+ "logits/rejected": -1.2823896408081055,
5348
+ "logps/chosen": -145.33433532714844,
5349
+ "logps/rejected": -154.925048828125,
5350
+ "loss": 0.53,
5351
+ "rewards/accuracies": 0.71875,
5352
+ "rewards/chosen": -0.1620834469795227,
5353
+ "rewards/margins": 0.8069512844085693,
5354
+ "rewards/rejected": -0.969034731388092,
5355
+ "step": 700
5356
+ },
5357
+ {
5358
+ "epoch": 0.8107129931234166,
5359
+ "eval_logits/chosen": -1.284436821937561,
5360
+ "eval_logits/rejected": -1.280255913734436,
5361
+ "eval_logps/chosen": -139.673828125,
5362
+ "eval_logps/rejected": -140.73330688476562,
5363
+ "eval_loss": 0.6434417963027954,
5364
+ "eval_rewards/accuracies": 0.6399999856948853,
5365
+ "eval_rewards/chosen": -0.4042653739452362,
5366
+ "eval_rewards/margins": 0.2737090587615967,
5367
+ "eval_rewards/rejected": -0.6779744625091553,
5368
+ "eval_runtime": 28.09,
5369
+ "eval_samples_per_second": 3.56,
5370
+ "eval_steps_per_second": 0.89,
5371
+ "step": 700
5372
+ },
5373
+ {
5374
+ "epoch": 0.8130293159609121,
5375
+ "grad_norm": 76.53295510100874,
5376
+ "learning_rate": 1.3964739704616247e-07,
5377
+ "logits/chosen": -1.2752941846847534,
5378
+ "logits/rejected": -1.3037452697753906,
5379
+ "logps/chosen": -121.41841125488281,
5380
+ "logps/rejected": -145.7820281982422,
5381
+ "loss": 0.5615,
5382
+ "rewards/accuracies": 0.71875,
5383
+ "rewards/chosen": -0.14446553587913513,
5384
+ "rewards/margins": 0.4738205671310425,
5385
+ "rewards/rejected": -0.6182860732078552,
5386
+ "step": 702
5387
+ },
5388
+ {
5389
+ "epoch": 0.8153456387984075,
5390
+ "grad_norm": 83.60642302801956,
5391
+ "learning_rate": 1.3929235082011233e-07,
5392
+ "logits/chosen": -1.296051025390625,
5393
+ "logits/rejected": -1.2584261894226074,
5394
+ "logps/chosen": -147.13363647460938,
5395
+ "logps/rejected": -150.47975158691406,
5396
+ "loss": 0.6121,
5397
+ "rewards/accuracies": 0.75,
5398
+ "rewards/chosen": -0.1639641672372818,
5399
+ "rewards/margins": 0.1966288536787033,
5400
+ "rewards/rejected": -0.3605930507183075,
5401
+ "step": 704
5402
+ },
5403
+ {
5404
+ "epoch": 0.817661961635903,
5405
+ "grad_norm": 80.9906935823332,
5406
+ "learning_rate": 1.3893671788050073e-07,
5407
+ "logits/chosen": -1.3390685319900513,
5408
+ "logits/rejected": -1.3612732887268066,
5409
+ "logps/chosen": -128.30557250976562,
5410
+ "logps/rejected": -151.2054443359375,
5411
+ "loss": 0.5976,
5412
+ "rewards/accuracies": 0.6875,
5413
+ "rewards/chosen": -0.28982260823249817,
5414
+ "rewards/margins": 0.5823266506195068,
5415
+ "rewards/rejected": -0.8721492886543274,
5416
+ "step": 706
5417
+ },
5418
+ {
5419
+ "epoch": 0.8199782844733985,
5420
+ "grad_norm": 79.56491935334587,
5421
+ "learning_rate": 1.385805035376403e-07,
5422
+ "logits/chosen": -1.2172850370407104,
5423
+ "logits/rejected": -1.2152363061904907,
5424
+ "logps/chosen": -127.4556884765625,
5425
+ "logps/rejected": -150.25283813476562,
5426
+ "loss": 0.6058,
5427
+ "rewards/accuracies": 0.71875,
5428
+ "rewards/chosen": -0.030050382018089294,
5429
+ "rewards/margins": 0.7187869548797607,
5430
+ "rewards/rejected": -0.7488372921943665,
5431
+ "step": 708
5432
+ },
5433
+ {
5434
+ "epoch": 0.822294607310894,
5435
+ "grad_norm": 94.61612804665803,
5436
+ "learning_rate": 1.3822371311052523e-07,
5437
+ "logits/chosen": -1.4023680686950684,
5438
+ "logits/rejected": -1.3874592781066895,
5439
+ "logps/chosen": -137.8409423828125,
5440
+ "logps/rejected": -148.14865112304688,
5441
+ "loss": 0.6506,
5442
+ "rewards/accuracies": 0.65625,
5443
+ "rewards/chosen": -0.055325593799352646,
5444
+ "rewards/margins": 0.21838925778865814,
5445
+ "rewards/rejected": -0.2737148702144623,
5446
+ "step": 710
5447
+ },
5448
+ {
5449
+ "epoch": 0.8246109301483895,
5450
+ "grad_norm": 128.27117224074266,
5451
+ "learning_rate": 1.3786635192675182e-07,
5452
+ "logits/chosen": -1.2412676811218262,
5453
+ "logits/rejected": -1.2453944683074951,
5454
+ "logps/chosen": -107.39035034179688,
5455
+ "logps/rejected": -125.48077392578125,
5456
+ "loss": 0.6535,
5457
+ "rewards/accuracies": 0.65625,
5458
+ "rewards/chosen": -0.47688642144203186,
5459
+ "rewards/margins": 0.8052099943161011,
5460
+ "rewards/rejected": -1.2820963859558105,
5461
+ "step": 712
5462
+ },
5463
+ {
5464
+ "epoch": 0.8269272529858849,
5465
+ "grad_norm": 81.45696425151283,
5466
+ "learning_rate": 1.375084253224389e-07,
5467
+ "logits/chosen": -1.2930258512496948,
5468
+ "logits/rejected": -1.275715708732605,
5469
+ "logps/chosen": -85.20941162109375,
5470
+ "logps/rejected": -94.45440673828125,
5471
+ "loss": 0.6138,
5472
+ "rewards/accuracies": 0.6875,
5473
+ "rewards/chosen": -0.1915649175643921,
5474
+ "rewards/margins": 0.5591136813163757,
5475
+ "rewards/rejected": -0.7506786584854126,
5476
+ "step": 714
5477
+ },
5478
+ {
5479
+ "epoch": 0.8292435758233804,
5480
+ "grad_norm": 68.28421801876017,
5481
+ "learning_rate": 1.371499386421481e-07,
5482
+ "logits/chosen": -1.4285151958465576,
5483
+ "logits/rejected": -1.4581302404403687,
5484
+ "logps/chosen": -144.27658081054688,
5485
+ "logps/rejected": -158.56509399414062,
5486
+ "loss": 0.5704,
5487
+ "rewards/accuracies": 0.8125,
5488
+ "rewards/chosen": 0.10951147228479385,
5489
+ "rewards/margins": 0.6330095529556274,
5490
+ "rewards/rejected": -0.5234980583190918,
5491
+ "step": 716
5492
+ },
5493
+ {
5494
+ "epoch": 0.8315598986608759,
5495
+ "grad_norm": 58.53704567635433,
5496
+ "learning_rate": 1.3679089723880426e-07,
5497
+ "logits/chosen": -1.2971611022949219,
5498
+ "logits/rejected": -1.349431037902832,
5499
+ "logps/chosen": -132.24440002441406,
5500
+ "logps/rejected": -157.0482940673828,
5501
+ "loss": 0.5487,
5502
+ "rewards/accuracies": 0.8125,
5503
+ "rewards/chosen": 0.03527238965034485,
5504
+ "rewards/margins": 0.778007447719574,
5505
+ "rewards/rejected": -0.7427350878715515,
5506
+ "step": 718
5507
+ },
5508
+ {
5509
+ "epoch": 0.8338762214983714,
5510
+ "grad_norm": 90.86882777075694,
5511
+ "learning_rate": 1.3643130647361515e-07,
5512
+ "logits/chosen": -1.223427414894104,
5513
+ "logits/rejected": -1.2997663021087646,
5514
+ "logps/chosen": -155.99490356445312,
5515
+ "logps/rejected": -184.7327880859375,
5516
+ "loss": 0.6357,
5517
+ "rewards/accuracies": 0.75,
5518
+ "rewards/chosen": -0.13655239343643188,
5519
+ "rewards/margins": 0.6724637746810913,
5520
+ "rewards/rejected": -0.8090161085128784,
5521
+ "step": 720
5522
+ },
5523
+ {
5524
+ "epoch": 0.8361925443358668,
5525
+ "grad_norm": 71.10479530353953,
5526
+ "learning_rate": 1.3607117171599178e-07,
5527
+ "logits/chosen": -1.432951807975769,
5528
+ "logits/rejected": -1.4361122846603394,
5529
+ "logps/chosen": -135.4175567626953,
5530
+ "logps/rejected": -137.8448944091797,
5531
+ "loss": 0.5987,
5532
+ "rewards/accuracies": 0.625,
5533
+ "rewards/chosen": -0.10024284571409225,
5534
+ "rewards/margins": 0.49338966608047485,
5535
+ "rewards/rejected": -0.5936326384544373,
5536
+ "step": 722
5537
+ },
5538
+ {
5539
+ "epoch": 0.8385088671733623,
5540
+ "grad_norm": 73.24864095064638,
5541
+ "learning_rate": 1.3571049834346796e-07,
5542
+ "logits/chosen": -1.266755223274231,
5543
+ "logits/rejected": -1.3649208545684814,
5544
+ "logps/chosen": -91.91627502441406,
5545
+ "logps/rejected": -137.7756805419922,
5546
+ "loss": 0.5877,
5547
+ "rewards/accuracies": 0.65625,
5548
+ "rewards/chosen": -0.03640535846352577,
5549
+ "rewards/margins": 0.47152870893478394,
5550
+ "rewards/rejected": -0.5079340934753418,
5551
+ "step": 724
5552
+ },
5553
+ {
5554
+ "epoch": 0.8408251900108578,
5555
+ "grad_norm": 84.42887169617501,
5556
+ "learning_rate": 1.3534929174162017e-07,
5557
+ "logits/chosen": -1.3389887809753418,
5558
+ "logits/rejected": -1.3608206510543823,
5559
+ "logps/chosen": -138.79879760742188,
5560
+ "logps/rejected": -148.29205322265625,
5561
+ "loss": 0.6315,
5562
+ "rewards/accuracies": 0.71875,
5563
+ "rewards/chosen": -0.13554176688194275,
5564
+ "rewards/margins": 0.4595597982406616,
5565
+ "rewards/rejected": -0.595101535320282,
5566
+ "step": 726
5567
+ },
5568
+ {
5569
+ "epoch": 0.8431415128483533,
5570
+ "grad_norm": 70.90955463261469,
5571
+ "learning_rate": 1.34987557303987e-07,
5572
+ "logits/chosen": -1.3609716892242432,
5573
+ "logits/rejected": -1.4318090677261353,
5574
+ "logps/chosen": -112.85796356201172,
5575
+ "logps/rejected": -138.29444885253906,
5576
+ "loss": 0.5358,
5577
+ "rewards/accuracies": 0.71875,
5578
+ "rewards/chosen": -0.0836462751030922,
5579
+ "rewards/margins": 0.7275322079658508,
5580
+ "rewards/rejected": -0.8111785650253296,
5581
+ "step": 728
5582
+ },
5583
+ {
5584
+ "epoch": 0.8454578356858488,
5585
+ "grad_norm": 114.31329319517884,
5586
+ "learning_rate": 1.3462530043198871e-07,
5587
+ "logits/chosen": -1.3260619640350342,
5588
+ "logits/rejected": -1.4074249267578125,
5589
+ "logps/chosen": -145.60401916503906,
5590
+ "logps/rejected": -165.5455780029297,
5591
+ "loss": 0.6309,
5592
+ "rewards/accuracies": 0.78125,
5593
+ "rewards/chosen": 0.12638582289218903,
5594
+ "rewards/margins": 0.7593765258789062,
5595
+ "rewards/rejected": -0.6329907178878784,
5596
+ "step": 730
5597
+ },
5598
+ {
5599
+ "epoch": 0.8477741585233441,
5600
+ "grad_norm": 109.74072899186363,
5601
+ "learning_rate": 1.342625265348466e-07,
5602
+ "logits/chosen": -1.3791258335113525,
5603
+ "logits/rejected": -1.4346139430999756,
5604
+ "logps/chosen": -178.01446533203125,
5605
+ "logps/rejected": -226.8712158203125,
5606
+ "loss": 0.6573,
5607
+ "rewards/accuracies": 0.65625,
5608
+ "rewards/chosen": -0.36265602707862854,
5609
+ "rewards/margins": 0.6506404280662537,
5610
+ "rewards/rejected": -1.0132964849472046,
5611
+ "step": 732
5612
+ },
5613
+ {
5614
+ "epoch": 0.8500904813608396,
5615
+ "grad_norm": 104.29186913107871,
5616
+ "learning_rate": 1.3389924102950213e-07,
5617
+ "logits/chosen": -1.3433104753494263,
5618
+ "logits/rejected": -1.349946141242981,
5619
+ "logps/chosen": -140.24569702148438,
5620
+ "logps/rejected": -149.5057373046875,
5621
+ "loss": 0.5876,
5622
+ "rewards/accuracies": 0.65625,
5623
+ "rewards/chosen": -0.04752161353826523,
5624
+ "rewards/margins": 0.3977040648460388,
5625
+ "rewards/rejected": -0.44522568583488464,
5626
+ "step": 734
5627
+ },
5628
+ {
5629
+ "epoch": 0.8524068041983351,
5630
+ "grad_norm": 81.62196698132477,
5631
+ "learning_rate": 1.3353544934053615e-07,
5632
+ "logits/chosen": -1.3260061740875244,
5633
+ "logits/rejected": -1.3429383039474487,
5634
+ "logps/chosen": -127.91857147216797,
5635
+ "logps/rejected": -150.63392639160156,
5636
+ "loss": 0.5993,
5637
+ "rewards/accuracies": 0.71875,
5638
+ "rewards/chosen": -0.27976539731025696,
5639
+ "rewards/margins": 0.31496569514274597,
5640
+ "rewards/rejected": -0.5947310924530029,
5641
+ "step": 736
5642
+ },
5643
+ {
5644
+ "epoch": 0.8547231270358306,
5645
+ "grad_norm": 88.51535267570014,
5646
+ "learning_rate": 1.331711569000878e-07,
5647
+ "logits/chosen": -1.385151982307434,
5648
+ "logits/rejected": -1.3968982696533203,
5649
+ "logps/chosen": -128.69894409179688,
5650
+ "logps/rejected": -145.05084228515625,
5651
+ "loss": 0.6054,
5652
+ "rewards/accuracies": 0.625,
5653
+ "rewards/chosen": 0.024006934836506844,
5654
+ "rewards/margins": 0.4703417718410492,
5655
+ "rewards/rejected": -0.4463347792625427,
5656
+ "step": 738
5657
+ },
5658
+ {
5659
+ "epoch": 0.8570394498733261,
5660
+ "grad_norm": 83.96836445474686,
5661
+ "learning_rate": 1.3280636914777344e-07,
5662
+ "logits/chosen": -1.2993488311767578,
5663
+ "logits/rejected": -1.3133535385131836,
5664
+ "logps/chosen": -133.5054168701172,
5665
+ "logps/rejected": -148.28396606445312,
5666
+ "loss": 0.557,
5667
+ "rewards/accuracies": 0.59375,
5668
+ "rewards/chosen": -0.11238986253738403,
5669
+ "rewards/margins": 0.47711944580078125,
5670
+ "rewards/rejected": -0.5895093083381653,
5671
+ "step": 740
5672
+ },
5673
+ {
5674
+ "epoch": 0.8593557727108215,
5675
+ "grad_norm": 83.73076557575334,
5676
+ "learning_rate": 1.3244109153060547e-07,
5677
+ "logits/chosen": -1.3348900079727173,
5678
+ "logits/rejected": -1.3360458612442017,
5679
+ "logps/chosen": -141.75746154785156,
5680
+ "logps/rejected": -149.81410217285156,
5681
+ "loss": 0.6232,
5682
+ "rewards/accuracies": 0.5625,
5683
+ "rewards/chosen": -0.05470200255513191,
5684
+ "rewards/margins": 0.1306239813566208,
5685
+ "rewards/rejected": -0.1853259950876236,
5686
+ "step": 742
5687
+ },
5688
+ {
5689
+ "epoch": 0.861672095548317,
5690
+ "grad_norm": 96.3207173875703,
5691
+ "learning_rate": 1.320753295029109e-07,
5692
+ "logits/chosen": -1.199431300163269,
5693
+ "logits/rejected": -1.2418042421340942,
5694
+ "logps/chosen": -101.57156372070312,
5695
+ "logps/rejected": -137.4820098876953,
5696
+ "loss": 0.6712,
5697
+ "rewards/accuracies": 0.71875,
5698
+ "rewards/chosen": -0.06123688444495201,
5699
+ "rewards/margins": 0.6316696405410767,
5700
+ "rewards/rejected": -0.6929064989089966,
5701
+ "step": 744
5702
+ },
5703
+ {
5704
+ "epoch": 0.8639884183858125,
5705
+ "grad_norm": 100.3243313129826,
5706
+ "learning_rate": 1.3170908852625012e-07,
5707
+ "logits/chosen": -1.3424568176269531,
5708
+ "logits/rejected": -1.3905185461044312,
5709
+ "logps/chosen": -149.32872009277344,
5710
+ "logps/rejected": -220.64122009277344,
5711
+ "loss": 0.5537,
5712
+ "rewards/accuracies": 0.71875,
5713
+ "rewards/chosen": -0.28289514780044556,
5714
+ "rewards/margins": 0.7176157236099243,
5715
+ "rewards/rejected": -1.000510811805725,
5716
+ "step": 746
5717
+ },
5718
+ {
5719
+ "epoch": 0.866304741223308,
5720
+ "grad_norm": 70.91956724923047,
5721
+ "learning_rate": 1.313423740693349e-07,
5722
+ "logits/chosen": -1.2025208473205566,
5723
+ "logits/rejected": -1.266943335533142,
5724
+ "logps/chosen": -152.86985778808594,
5725
+ "logps/rejected": -173.8192138671875,
5726
+ "loss": 0.6223,
5727
+ "rewards/accuracies": 0.5625,
5728
+ "rewards/chosen": -0.04741968959569931,
5729
+ "rewards/margins": 0.35572314262390137,
5730
+ "rewards/rejected": -0.4031428098678589,
5731
+ "step": 748
5732
+ },
5733
+ {
5734
+ "epoch": 0.8686210640608035,
5735
+ "grad_norm": 75.21549346823406,
5736
+ "learning_rate": 1.3097519160794723e-07,
5737
+ "logits/chosen": -1.3221216201782227,
5738
+ "logits/rejected": -1.326745629310608,
5739
+ "logps/chosen": -105.75431823730469,
5740
+ "logps/rejected": -113.87251281738281,
5741
+ "loss": 0.5687,
5742
+ "rewards/accuracies": 0.78125,
5743
+ "rewards/chosen": -0.058674052357673645,
5744
+ "rewards/margins": 0.6489120125770569,
5745
+ "rewards/rejected": -0.7075860500335693,
5746
+ "step": 750
5747
+ },
5748
+ {
5749
+ "epoch": 0.8709373868982989,
5750
+ "grad_norm": 74.6437050803779,
5751
+ "learning_rate": 1.306075466248574e-07,
5752
+ "logits/chosen": -1.3908584117889404,
5753
+ "logits/rejected": -1.3725515604019165,
5754
+ "logps/chosen": -156.27899169921875,
5755
+ "logps/rejected": -151.42625427246094,
5756
+ "loss": 0.5536,
5757
+ "rewards/accuracies": 0.71875,
5758
+ "rewards/chosen": 0.030293064191937447,
5759
+ "rewards/margins": 0.33356091380119324,
5760
+ "rewards/rejected": -0.30326780676841736,
5761
+ "step": 752
5762
+ },
5763
+ {
5764
+ "epoch": 0.8732537097357944,
5765
+ "grad_norm": 86.78294546912407,
5766
+ "learning_rate": 1.302394446097418e-07,
5767
+ "logits/chosen": -1.3920029401779175,
5768
+ "logits/rejected": -1.4190560579299927,
5769
+ "logps/chosen": -131.45228576660156,
5770
+ "logps/rejected": -169.52101135253906,
5771
+ "loss": 0.6078,
5772
+ "rewards/accuracies": 0.59375,
5773
+ "rewards/chosen": -0.09971225261688232,
5774
+ "rewards/margins": 0.5380171537399292,
5775
+ "rewards/rejected": -0.6377294659614563,
5776
+ "step": 754
5777
+ },
5778
+ {
5779
+ "epoch": 0.8755700325732899,
5780
+ "grad_norm": 74.55059453725265,
5781
+ "learning_rate": 1.2987089105910155e-07,
5782
+ "logits/chosen": -1.239861011505127,
5783
+ "logits/rejected": -1.2780354022979736,
5784
+ "logps/chosen": -157.7389373779297,
5785
+ "logps/rejected": -199.31524658203125,
5786
+ "loss": 0.5538,
5787
+ "rewards/accuracies": 0.78125,
5788
+ "rewards/chosen": 0.019644632935523987,
5789
+ "rewards/margins": 1.3704544305801392,
5790
+ "rewards/rejected": -1.3508098125457764,
5791
+ "step": 756
5792
+ },
5793
+ {
5794
+ "epoch": 0.8778863554107854,
5795
+ "grad_norm": 76.3646913429212,
5796
+ "learning_rate": 1.2950189147617987e-07,
5797
+ "logits/chosen": -1.484637975692749,
5798
+ "logits/rejected": -1.451521873474121,
5799
+ "logps/chosen": -216.90333557128906,
5800
+ "logps/rejected": -216.04563903808594,
5801
+ "loss": 0.5957,
5802
+ "rewards/accuracies": 0.625,
5803
+ "rewards/chosen": -0.4405128061771393,
5804
+ "rewards/margins": 0.5718247294425964,
5805
+ "rewards/rejected": -1.012337565422058,
5806
+ "step": 758
5807
+ },
5808
+ {
5809
+ "epoch": 0.8802026782482808,
5810
+ "grad_norm": 81.06900576809731,
5811
+ "learning_rate": 1.2913245137088022e-07,
5812
+ "logits/chosen": -1.4075809717178345,
5813
+ "logits/rejected": -1.446765422821045,
5814
+ "logps/chosen": -182.1182861328125,
5815
+ "logps/rejected": -211.40109252929688,
5816
+ "loss": 0.5711,
5817
+ "rewards/accuracies": 0.71875,
5818
+ "rewards/chosen": -0.01186487078666687,
5819
+ "rewards/margins": 0.6298959255218506,
5820
+ "rewards/rejected": -0.6417607665061951,
5821
+ "step": 760
5822
+ },
5823
+ {
5824
+ "epoch": 0.8825190010857763,
5825
+ "grad_norm": 105.35884658750386,
5826
+ "learning_rate": 1.2876257625968397e-07,
5827
+ "logits/chosen": -1.2835676670074463,
5828
+ "logits/rejected": -1.3046995401382446,
5829
+ "logps/chosen": -132.78152465820312,
5830
+ "logps/rejected": -161.31594848632812,
5831
+ "loss": 0.5776,
5832
+ "rewards/accuracies": 0.78125,
5833
+ "rewards/chosen": -0.09922360628843307,
5834
+ "rewards/margins": 0.8614255785942078,
5835
+ "rewards/rejected": -0.9606491327285767,
5836
+ "step": 762
5837
+ },
5838
+ {
5839
+ "epoch": 0.8848353239232718,
5840
+ "grad_norm": 79.20062356379438,
5841
+ "learning_rate": 1.283922716655679e-07,
5842
+ "logits/chosen": -1.1998169422149658,
5843
+ "logits/rejected": -1.2543299198150635,
5844
+ "logps/chosen": -123.684326171875,
5845
+ "logps/rejected": -152.342041015625,
5846
+ "loss": 0.5882,
5847
+ "rewards/accuracies": 0.6875,
5848
+ "rewards/chosen": -0.04377639666199684,
5849
+ "rewards/margins": 0.5460759997367859,
5850
+ "rewards/rejected": -0.5898523926734924,
5851
+ "step": 764
5852
+ },
5853
+ {
5854
+ "epoch": 0.8871516467607673,
5855
+ "grad_norm": 83.51961253986426,
5856
+ "learning_rate": 1.2802154311792196e-07,
5857
+ "logits/chosen": -1.1967344284057617,
5858
+ "logits/rejected": -1.2713446617126465,
5859
+ "logps/chosen": -139.49871826171875,
5860
+ "logps/rejected": -159.91653442382812,
5861
+ "loss": 0.609,
5862
+ "rewards/accuracies": 0.65625,
5863
+ "rewards/chosen": -0.07481690496206284,
5864
+ "rewards/margins": 0.5285822153091431,
5865
+ "rewards/rejected": -0.6033991575241089,
5866
+ "step": 766
5867
+ },
5868
+ {
5869
+ "epoch": 0.8894679695982628,
5870
+ "grad_norm": 87.62676156964712,
5871
+ "learning_rate": 1.276503961524665e-07,
5872
+ "logits/chosen": -1.3275455236434937,
5873
+ "logits/rejected": -1.4026497602462769,
5874
+ "logps/chosen": -129.25172424316406,
5875
+ "logps/rejected": -143.60403442382812,
5876
+ "loss": 0.6153,
5877
+ "rewards/accuracies": 0.6875,
5878
+ "rewards/chosen": -0.25011736154556274,
5879
+ "rewards/margins": 0.5108656883239746,
5880
+ "rewards/rejected": -0.7609830498695374,
5881
+ "step": 768
5882
+ },
5883
+ {
5884
+ "epoch": 0.8917842924357582,
5885
+ "grad_norm": 79.69384964872866,
5886
+ "learning_rate": 1.2727883631116967e-07,
5887
+ "logits/chosen": -1.250293493270874,
5888
+ "logits/rejected": -1.371201753616333,
5889
+ "logps/chosen": -165.1368865966797,
5890
+ "logps/rejected": -193.74635314941406,
5891
+ "loss": 0.5857,
5892
+ "rewards/accuracies": 0.59375,
5893
+ "rewards/chosen": -0.06070555001497269,
5894
+ "rewards/margins": 0.8822014331817627,
5895
+ "rewards/rejected": -0.94290691614151,
5896
+ "step": 770
5897
+ },
5898
+ {
5899
+ "epoch": 0.8941006152732537,
5900
+ "grad_norm": 208.13208814234412,
5901
+ "learning_rate": 1.2690686914216473e-07,
5902
+ "logits/chosen": -1.389907956123352,
5903
+ "logits/rejected": -1.372930645942688,
5904
+ "logps/chosen": -154.89971923828125,
5905
+ "logps/rejected": -180.29736328125,
5906
+ "loss": 0.7413,
5907
+ "rewards/accuracies": 0.53125,
5908
+ "rewards/chosen": -0.334947407245636,
5909
+ "rewards/margins": 0.8569056391716003,
5910
+ "rewards/rejected": -1.1918531656265259,
5911
+ "step": 772
5912
+ },
5913
+ {
5914
+ "epoch": 0.8964169381107492,
5915
+ "grad_norm": 87.24757168930763,
5916
+ "learning_rate": 1.2653450019966719e-07,
5917
+ "logits/chosen": -1.2602267265319824,
5918
+ "logits/rejected": -1.2742539644241333,
5919
+ "logps/chosen": -108.18531036376953,
5920
+ "logps/rejected": -128.6900634765625,
5921
+ "loss": 0.5774,
5922
+ "rewards/accuracies": 0.59375,
5923
+ "rewards/chosen": -0.06359031796455383,
5924
+ "rewards/margins": 0.7088490128517151,
5925
+ "rewards/rejected": -0.7724392414093018,
5926
+ "step": 774
5927
+ },
5928
+ {
5929
+ "epoch": 0.8987332609482447,
5930
+ "grad_norm": 83.80268761754905,
5931
+ "learning_rate": 1.2616173504389172e-07,
5932
+ "logits/chosen": -1.2021056413650513,
5933
+ "logits/rejected": -1.2368252277374268,
5934
+ "logps/chosen": -141.87701416015625,
5935
+ "logps/rejected": -154.36065673828125,
5936
+ "loss": 0.6236,
5937
+ "rewards/accuracies": 0.5,
5938
+ "rewards/chosen": -0.3400351405143738,
5939
+ "rewards/margins": 0.41063636541366577,
5940
+ "rewards/rejected": -0.7506715655326843,
5941
+ "step": 776
5942
+ },
5943
+ {
5944
+ "epoch": 0.9010495837857402,
5945
+ "grad_norm": 153.2532920637624,
5946
+ "learning_rate": 1.2578857924096933e-07,
5947
+ "logits/chosen": -1.448024034500122,
5948
+ "logits/rejected": -1.3988113403320312,
5949
+ "logps/chosen": -154.5913848876953,
5950
+ "logps/rejected": -158.92227172851562,
5951
+ "loss": 0.6673,
5952
+ "rewards/accuracies": 0.6875,
5953
+ "rewards/chosen": -0.05520397424697876,
5954
+ "rewards/margins": 0.35100066661834717,
5955
+ "rewards/rejected": -0.4062046408653259,
5956
+ "step": 778
5957
+ },
5958
+ {
5959
+ "epoch": 0.9033659066232356,
5960
+ "grad_norm": 71.28047852511637,
5961
+ "learning_rate": 1.2541503836286425e-07,
5962
+ "logits/chosen": -1.3942607641220093,
5963
+ "logits/rejected": -1.349011778831482,
5964
+ "logps/chosen": -173.88368225097656,
5965
+ "logps/rejected": -171.93663024902344,
5966
+ "loss": 0.5815,
5967
+ "rewards/accuracies": 0.84375,
5968
+ "rewards/chosen": -0.220017209649086,
5969
+ "rewards/margins": 0.7204620242118835,
5970
+ "rewards/rejected": -0.9404792785644531,
5971
+ "step": 780
5972
+ },
5973
+ {
5974
+ "epoch": 0.9056822294607311,
5975
+ "grad_norm": 111.86020202854063,
5976
+ "learning_rate": 1.250411179872905e-07,
5977
+ "logits/chosen": -1.2979280948638916,
5978
+ "logits/rejected": -1.3246498107910156,
5979
+ "logps/chosen": -132.1781463623047,
5980
+ "logps/rejected": -142.88433837890625,
5981
+ "loss": 0.5809,
5982
+ "rewards/accuracies": 0.6875,
5983
+ "rewards/chosen": -0.1429004967212677,
5984
+ "rewards/margins": 0.4971502721309662,
5985
+ "rewards/rejected": -0.6400507688522339,
5986
+ "step": 782
5987
+ },
5988
+ {
5989
+ "epoch": 0.9079985522982266,
5990
+ "grad_norm": 72.6068979409082,
5991
+ "learning_rate": 1.246668236976288e-07,
5992
+ "logits/chosen": -1.365678310394287,
5993
+ "logits/rejected": -1.2877997159957886,
5994
+ "logps/chosen": -132.5888671875,
5995
+ "logps/rejected": -125.15145874023438,
5996
+ "loss": 0.6167,
5997
+ "rewards/accuracies": 0.5625,
5998
+ "rewards/chosen": -0.05012892931699753,
5999
+ "rewards/margins": 0.14901477098464966,
6000
+ "rewards/rejected": -0.19914370775222778,
6001
+ "step": 784
6002
+ },
6003
+ {
6004
+ "epoch": 0.9103148751357221,
6005
+ "grad_norm": 77.28231066519794,
6006
+ "learning_rate": 1.2429216108284333e-07,
6007
+ "logits/chosen": -1.241924524307251,
6008
+ "logits/rejected": -1.2669358253479004,
6009
+ "logps/chosen": -170.2476348876953,
6010
+ "logps/rejected": -176.51695251464844,
6011
+ "loss": 0.6174,
6012
+ "rewards/accuracies": 0.6875,
6013
+ "rewards/chosen": -0.2486531287431717,
6014
+ "rewards/margins": 0.7952675223350525,
6015
+ "rewards/rejected": -1.0439205169677734,
6016
+ "step": 786
6017
+ },
6018
+ {
6019
+ "epoch": 0.9126311979732176,
6020
+ "grad_norm": 77.68314015122489,
6021
+ "learning_rate": 1.2391713573739784e-07,
6022
+ "logits/chosen": -1.5044151544570923,
6023
+ "logits/rejected": -1.38852059841156,
6024
+ "logps/chosen": -156.23318481445312,
6025
+ "logps/rejected": -150.65298461914062,
6026
+ "loss": 0.5763,
6027
+ "rewards/accuracies": 0.5,
6028
+ "rewards/chosen": -0.028940770775079727,
6029
+ "rewards/margins": 0.28721868991851807,
6030
+ "rewards/rejected": -0.3161594271659851,
6031
+ "step": 788
6032
+ },
6033
+ {
6034
+ "epoch": 0.914947520810713,
6035
+ "grad_norm": 74.30222390288303,
6036
+ "learning_rate": 1.235417532611725e-07,
6037
+ "logits/chosen": -1.4984527826309204,
6038
+ "logits/rejected": -1.5665603876113892,
6039
+ "logps/chosen": -166.45755004882812,
6040
+ "logps/rejected": -199.21507263183594,
6041
+ "loss": 0.6066,
6042
+ "rewards/accuracies": 0.65625,
6043
+ "rewards/chosen": -0.00897267460823059,
6044
+ "rewards/margins": 0.5808253288269043,
6045
+ "rewards/rejected": -0.5897979736328125,
6046
+ "step": 790
6047
+ },
6048
+ {
6049
+ "epoch": 0.9172638436482085,
6050
+ "grad_norm": 73.07953128202571,
6051
+ "learning_rate": 1.2316601925938025e-07,
6052
+ "logits/chosen": -1.3732727766036987,
6053
+ "logits/rejected": -1.444867730140686,
6054
+ "logps/chosen": -126.97662353515625,
6055
+ "logps/rejected": -142.06971740722656,
6056
+ "loss": 0.5687,
6057
+ "rewards/accuracies": 0.65625,
6058
+ "rewards/chosen": -0.05256509408354759,
6059
+ "rewards/margins": 0.28611257672309875,
6060
+ "rewards/rejected": -0.33867767453193665,
6061
+ "step": 792
6062
+ },
6063
+ {
6064
+ "epoch": 0.919580166485704,
6065
+ "grad_norm": 95.48085015281728,
6066
+ "learning_rate": 1.2278993934248278e-07,
6067
+ "logits/chosen": -1.258427381515503,
6068
+ "logits/rejected": -1.275139570236206,
6069
+ "logps/chosen": -110.3523178100586,
6070
+ "logps/rejected": -123.8600082397461,
6071
+ "loss": 0.692,
6072
+ "rewards/accuracies": 0.6875,
6073
+ "rewards/chosen": -0.11288494616746902,
6074
+ "rewards/margins": 0.2848440706729889,
6075
+ "rewards/rejected": -0.3977290093898773,
6076
+ "step": 794
6077
+ },
6078
+ {
6079
+ "epoch": 0.9218964893231995,
6080
+ "grad_norm": 58.55738169440391,
6081
+ "learning_rate": 1.2241351912610725e-07,
6082
+ "logits/chosen": -1.2354912757873535,
6083
+ "logits/rejected": -1.2666208744049072,
6084
+ "logps/chosen": -107.17646026611328,
6085
+ "logps/rejected": -116.17412567138672,
6086
+ "loss": 0.6002,
6087
+ "rewards/accuracies": 0.59375,
6088
+ "rewards/chosen": -0.3244227468967438,
6089
+ "rewards/margins": 0.10813488066196442,
6090
+ "rewards/rejected": -0.4325576424598694,
6091
+ "step": 796
6092
+ },
6093
+ {
6094
+ "epoch": 0.924212812160695,
6095
+ "grad_norm": 79.9120853192708,
6096
+ "learning_rate": 1.2203676423096197e-07,
6097
+ "logits/chosen": -1.2370004653930664,
6098
+ "logits/rejected": -1.348116159439087,
6099
+ "logps/chosen": -173.00152587890625,
6100
+ "logps/rejected": -202.92649841308594,
6101
+ "loss": 0.5644,
6102
+ "rewards/accuracies": 0.71875,
6103
+ "rewards/chosen": -0.16612032055854797,
6104
+ "rewards/margins": 0.47205114364624023,
6105
+ "rewards/rejected": -0.6381714344024658,
6106
+ "step": 798
6107
+ },
6108
+ {
6109
+ "epoch": 0.9265291349981903,
6110
+ "grad_norm": 70.53190608979963,
6111
+ "learning_rate": 1.2165968028275275e-07,
6112
+ "logits/chosen": -1.3343358039855957,
6113
+ "logits/rejected": -1.339569330215454,
6114
+ "logps/chosen": -177.5317840576172,
6115
+ "logps/rejected": -194.53817749023438,
6116
+ "loss": 0.5761,
6117
+ "rewards/accuracies": 0.625,
6118
+ "rewards/chosen": -0.04334399104118347,
6119
+ "rewards/margins": 0.66258704662323,
6120
+ "rewards/rejected": -0.7059310078620911,
6121
+ "step": 800
6122
+ },
6123
+ {
6124
+ "epoch": 0.9265291349981903,
6125
+ "eval_logits/chosen": -1.319779396057129,
6126
+ "eval_logits/rejected": -1.3124960660934448,
6127
+ "eval_logps/chosen": -139.39353942871094,
6128
+ "eval_logps/rejected": -140.94285583496094,
6129
+ "eval_loss": 0.6185734272003174,
6130
+ "eval_rewards/accuracies": 0.6399999856948853,
6131
+ "eval_rewards/chosen": -0.37623491883277893,
6132
+ "eval_rewards/margins": 0.32269400358200073,
6133
+ "eval_rewards/rejected": -0.6989290118217468,
6134
+ "eval_runtime": 29.0213,
6135
+ "eval_samples_per_second": 3.446,
6136
+ "eval_steps_per_second": 0.861,
6137
+ "step": 800
6138
+ },
6139
+ {
6140
+ "epoch": 0.9288454578356858,
6141
+ "grad_norm": 80.16286794415615,
6142
+ "learning_rate": 1.212822729120989e-07,
6143
+ "logits/chosen": -1.4367756843566895,
6144
+ "logits/rejected": -1.3805702924728394,
6145
+ "logps/chosen": -171.19290161132812,
6146
+ "logps/rejected": -173.32327270507812,
6147
+ "loss": 0.6077,
6148
+ "rewards/accuracies": 0.625,
6149
+ "rewards/chosen": -0.2906244695186615,
6150
+ "rewards/margins": 0.5511162281036377,
6151
+ "rewards/rejected": -0.8417407274246216,
6152
+ "step": 802
6153
+ },
6154
+ {
6155
+ "epoch": 0.9311617806731813,
6156
+ "grad_norm": 73.70947981864272,
6157
+ "learning_rate": 1.209045477544489e-07,
6158
+ "logits/chosen": -1.219347357749939,
6159
+ "logits/rejected": -1.3135026693344116,
6160
+ "logps/chosen": -125.98372650146484,
6161
+ "logps/rejected": -145.56602478027344,
6162
+ "loss": 0.5605,
6163
+ "rewards/accuracies": 0.75,
6164
+ "rewards/chosen": -0.173287034034729,
6165
+ "rewards/margins": 0.5581431984901428,
6166
+ "rewards/rejected": -0.7314302325248718,
6167
+ "step": 804
6168
+ },
6169
+ {
6170
+ "epoch": 0.9334781035106768,
6171
+ "grad_norm": 103.71883284173158,
6172
+ "learning_rate": 1.2052651044999658e-07,
6173
+ "logits/chosen": -1.2322499752044678,
6174
+ "logits/rejected": -1.3243571519851685,
6175
+ "logps/chosen": -103.25643157958984,
6176
+ "logps/rejected": -140.8385009765625,
6177
+ "loss": 0.5835,
6178
+ "rewards/accuracies": 0.65625,
6179
+ "rewards/chosen": -0.15573744475841522,
6180
+ "rewards/margins": 0.8674708008766174,
6181
+ "rewards/rejected": -1.0232083797454834,
6182
+ "step": 806
6183
+ },
6184
+ {
6185
+ "epoch": 0.9357944263481722,
6186
+ "grad_norm": 62.392714879995744,
6187
+ "learning_rate": 1.201481666435967e-07,
6188
+ "logits/chosen": -1.3585785627365112,
6189
+ "logits/rejected": -1.3998768329620361,
6190
+ "logps/chosen": -153.4759521484375,
6191
+ "logps/rejected": -162.12579345703125,
6192
+ "loss": 0.5661,
6193
+ "rewards/accuracies": 0.6875,
6194
+ "rewards/chosen": -0.23147635161876678,
6195
+ "rewards/margins": 0.3873007893562317,
6196
+ "rewards/rejected": -0.6187771558761597,
6197
+ "step": 808
6198
+ },
6199
+ {
6200
+ "epoch": 0.9381107491856677,
6201
+ "grad_norm": 92.78640979629462,
6202
+ "learning_rate": 1.1976952198468065e-07,
6203
+ "logits/chosen": -1.3264836072921753,
6204
+ "logits/rejected": -1.3320363759994507,
6205
+ "logps/chosen": -168.32054138183594,
6206
+ "logps/rejected": -156.73651123046875,
6207
+ "loss": 0.5762,
6208
+ "rewards/accuracies": 0.71875,
6209
+ "rewards/chosen": -0.20424926280975342,
6210
+ "rewards/margins": 0.26160287857055664,
6211
+ "rewards/rejected": -0.46585214138031006,
6212
+ "step": 810
6213
+ },
6214
+ {
6215
+ "epoch": 0.9404270720231632,
6216
+ "grad_norm": 81.23624930191687,
6217
+ "learning_rate": 1.1939058212717224e-07,
6218
+ "logits/chosen": -1.25242280960083,
6219
+ "logits/rejected": -1.2761772871017456,
6220
+ "logps/chosen": -142.50938415527344,
6221
+ "logps/rejected": -162.35137939453125,
6222
+ "loss": 0.6122,
6223
+ "rewards/accuracies": 0.59375,
6224
+ "rewards/chosen": -0.40755337476730347,
6225
+ "rewards/margins": 0.8417052030563354,
6226
+ "rewards/rejected": -1.2492586374282837,
6227
+ "step": 812
6228
+ },
6229
+ {
6230
+ "epoch": 0.9427433948606587,
6231
+ "grad_norm": 93.63860671562202,
6232
+ "learning_rate": 1.1901135272940319e-07,
6233
+ "logits/chosen": -1.3451721668243408,
6234
+ "logits/rejected": -1.4240652322769165,
6235
+ "logps/chosen": -166.04376220703125,
6236
+ "logps/rejected": -201.26220703125,
6237
+ "loss": 0.5927,
6238
+ "rewards/accuracies": 0.75,
6239
+ "rewards/chosen": -0.3375451862812042,
6240
+ "rewards/margins": 0.7751595377922058,
6241
+ "rewards/rejected": -1.112704873085022,
6242
+ "step": 814
6243
+ },
6244
+ {
6245
+ "epoch": 0.9450597176981542,
6246
+ "grad_norm": 74.49560053095384,
6247
+ "learning_rate": 1.1863183945402853e-07,
6248
+ "logits/chosen": -1.399809718132019,
6249
+ "logits/rejected": -1.4244718551635742,
6250
+ "logps/chosen": -142.70849609375,
6251
+ "logps/rejected": -154.3977508544922,
6252
+ "loss": 0.6206,
6253
+ "rewards/accuracies": 0.59375,
6254
+ "rewards/chosen": -0.27791112661361694,
6255
+ "rewards/margins": 0.23856940865516663,
6256
+ "rewards/rejected": -0.5164804458618164,
6257
+ "step": 816
6258
+ },
6259
+ {
6260
+ "epoch": 0.9473760405356496,
6261
+ "grad_norm": 76.5360836871227,
6262
+ "learning_rate": 1.1825204796794222e-07,
6263
+ "logits/chosen": -1.2848563194274902,
6264
+ "logits/rejected": -1.3460373878479004,
6265
+ "logps/chosen": -126.32781982421875,
6266
+ "logps/rejected": -146.41732788085938,
6267
+ "loss": 0.5922,
6268
+ "rewards/accuracies": 0.65625,
6269
+ "rewards/chosen": -0.04460408538579941,
6270
+ "rewards/margins": 0.6903345584869385,
6271
+ "rewards/rejected": -0.7349386215209961,
6272
+ "step": 818
6273
+ },
6274
+ {
6275
+ "epoch": 0.9496923633731451,
6276
+ "grad_norm": 69.37848187371887,
6277
+ "learning_rate": 1.1787198394219248e-07,
6278
+ "logits/chosen": -1.212153434753418,
6279
+ "logits/rejected": -1.2568764686584473,
6280
+ "logps/chosen": -112.97404479980469,
6281
+ "logps/rejected": -127.22880554199219,
6282
+ "loss": 0.5922,
6283
+ "rewards/accuracies": 0.65625,
6284
+ "rewards/chosen": -0.2542499303817749,
6285
+ "rewards/margins": 0.43802982568740845,
6286
+ "rewards/rejected": -0.6922798156738281,
6287
+ "step": 820
6288
+ },
6289
+ {
6290
+ "epoch": 0.9520086862106406,
6291
+ "grad_norm": 74.78466523725046,
6292
+ "learning_rate": 1.1749165305189708e-07,
6293
+ "logits/chosen": -1.2366949319839478,
6294
+ "logits/rejected": -1.3136887550354004,
6295
+ "logps/chosen": -135.05006408691406,
6296
+ "logps/rejected": -167.1379852294922,
6297
+ "loss": 0.6245,
6298
+ "rewards/accuracies": 0.875,
6299
+ "rewards/chosen": -0.1665336787700653,
6300
+ "rewards/margins": 0.8222386837005615,
6301
+ "rewards/rejected": -0.9887723922729492,
6302
+ "step": 822
6303
+ },
6304
+ {
6305
+ "epoch": 0.9543250090481361,
6306
+ "grad_norm": 114.72850198751108,
6307
+ "learning_rate": 1.1711106097615862e-07,
6308
+ "logits/chosen": -1.2289320230484009,
6309
+ "logits/rejected": -1.2598837614059448,
6310
+ "logps/chosen": -175.93649291992188,
6311
+ "logps/rejected": -174.3079376220703,
6312
+ "loss": 0.7136,
6313
+ "rewards/accuracies": 0.4375,
6314
+ "rewards/chosen": -0.8970119953155518,
6315
+ "rewards/margins": -0.25789961218833923,
6316
+ "rewards/rejected": -0.6391124725341797,
6317
+ "step": 824
6318
+ },
6319
+ {
6320
+ "epoch": 0.9566413318856316,
6321
+ "grad_norm": 82.7729985201964,
6322
+ "learning_rate": 1.1673021339797967e-07,
6323
+ "logits/chosen": -1.3273519277572632,
6324
+ "logits/rejected": -1.309187412261963,
6325
+ "logps/chosen": -123.46469116210938,
6326
+ "logps/rejected": -143.95350646972656,
6327
+ "loss": 0.6453,
6328
+ "rewards/accuracies": 0.75,
6329
+ "rewards/chosen": -0.04680802673101425,
6330
+ "rewards/margins": 0.8337461352348328,
6331
+ "rewards/rejected": -0.88055419921875,
6332
+ "step": 826
6333
+ },
6334
+ {
6335
+ "epoch": 0.958957654723127,
6336
+ "grad_norm": 93.78229578972356,
6337
+ "learning_rate": 1.16349116004178e-07,
6338
+ "logits/chosen": -1.235456943511963,
6339
+ "logits/rejected": -1.2821218967437744,
6340
+ "logps/chosen": -108.3389892578125,
6341
+ "logps/rejected": -153.83152770996094,
6342
+ "loss": 0.5369,
6343
+ "rewards/accuracies": 0.65625,
6344
+ "rewards/chosen": -0.07939951866865158,
6345
+ "rewards/margins": 0.8632293939590454,
6346
+ "rewards/rejected": -0.942628800868988,
6347
+ "step": 828
6348
+ },
6349
+ {
6350
+ "epoch": 0.9612739775606225,
6351
+ "grad_norm": 82.41065568509865,
6352
+ "learning_rate": 1.1596777448530169e-07,
6353
+ "logits/chosen": -1.2206637859344482,
6354
+ "logits/rejected": -1.2953643798828125,
6355
+ "logps/chosen": -127.601318359375,
6356
+ "logps/rejected": -152.67910766601562,
6357
+ "loss": 0.569,
6358
+ "rewards/accuracies": 0.75,
6359
+ "rewards/chosen": 0.004549721255898476,
6360
+ "rewards/margins": 1.1071059703826904,
6361
+ "rewards/rejected": -1.1025562286376953,
6362
+ "step": 830
6363
+ },
6364
+ {
6365
+ "epoch": 0.963590300398118,
6366
+ "grad_norm": 73.37771598718861,
6367
+ "learning_rate": 1.15586194535544e-07,
6368
+ "logits/chosen": -1.3030959367752075,
6369
+ "logits/rejected": -1.3053159713745117,
6370
+ "logps/chosen": -145.50299072265625,
6371
+ "logps/rejected": -202.24827575683594,
6372
+ "loss": 0.5484,
6373
+ "rewards/accuracies": 0.71875,
6374
+ "rewards/chosen": -0.16388273239135742,
6375
+ "rewards/margins": 2.732254981994629,
6376
+ "rewards/rejected": -2.8961377143859863,
6377
+ "step": 832
6378
+ },
6379
+ {
6380
+ "epoch": 0.9659066232356135,
6381
+ "grad_norm": 77.35826158650924,
6382
+ "learning_rate": 1.1520438185265846e-07,
6383
+ "logits/chosen": -1.3323960304260254,
6384
+ "logits/rejected": -1.3863334655761719,
6385
+ "logps/chosen": -129.73374938964844,
6386
+ "logps/rejected": -150.87632751464844,
6387
+ "loss": 0.6108,
6388
+ "rewards/accuracies": 0.625,
6389
+ "rewards/chosen": -0.013788050971925259,
6390
+ "rewards/margins": 0.6643078923225403,
6391
+ "rewards/rejected": -0.6780959367752075,
6392
+ "step": 834
6393
+ },
6394
+ {
6395
+ "epoch": 0.968222946073109,
6396
+ "grad_norm": 71.75741636541773,
6397
+ "learning_rate": 1.1482234213787383e-07,
6398
+ "logits/chosen": -1.404686689376831,
6399
+ "logits/rejected": -1.4628666639328003,
6400
+ "logps/chosen": -105.55296325683594,
6401
+ "logps/rejected": -159.8828582763672,
6402
+ "loss": 0.5356,
6403
+ "rewards/accuracies": 0.8125,
6404
+ "rewards/chosen": 0.02817944809794426,
6405
+ "rewards/margins": 0.7608631253242493,
6406
+ "rewards/rejected": -0.7326837182044983,
6407
+ "step": 836
6408
+ },
6409
+ {
6410
+ "epoch": 0.9705392689106044,
6411
+ "grad_norm": 70.54976379165657,
6412
+ "learning_rate": 1.1444008109580882e-07,
6413
+ "logits/chosen": -1.1081938743591309,
6414
+ "logits/rejected": -1.19542396068573,
6415
+ "logps/chosen": -109.79875946044922,
6416
+ "logps/rejected": -166.45217895507812,
6417
+ "loss": 0.5609,
6418
+ "rewards/accuracies": 0.75,
6419
+ "rewards/chosen": -0.2566208839416504,
6420
+ "rewards/margins": 0.9902852773666382,
6421
+ "rewards/rejected": -1.2469063997268677,
6422
+ "step": 838
6423
+ },
6424
+ {
6425
+ "epoch": 0.9728555917480999,
6426
+ "grad_norm": 68.85449968110935,
6427
+ "learning_rate": 1.1405760443438712e-07,
6428
+ "logits/chosen": -1.4356677532196045,
6429
+ "logits/rejected": -1.4843618869781494,
6430
+ "logps/chosen": -160.30462646484375,
6431
+ "logps/rejected": -174.40020751953125,
6432
+ "loss": 0.5569,
6433
+ "rewards/accuracies": 0.8125,
6434
+ "rewards/chosen": -0.0007999027147889137,
6435
+ "rewards/margins": 0.5300293564796448,
6436
+ "rewards/rejected": -0.5308292508125305,
6437
+ "step": 840
6438
+ },
6439
+ {
6440
+ "epoch": 0.9751719145855954,
6441
+ "grad_norm": 87.62002715815144,
6442
+ "learning_rate": 1.1367491786475194e-07,
6443
+ "logits/chosen": -1.3982141017913818,
6444
+ "logits/rejected": -1.436221957206726,
6445
+ "logps/chosen": -142.02593994140625,
6446
+ "logps/rejected": -171.52378845214844,
6447
+ "loss": 0.5838,
6448
+ "rewards/accuracies": 0.65625,
6449
+ "rewards/chosen": -0.1319505125284195,
6450
+ "rewards/margins": 0.5477418303489685,
6451
+ "rewards/rejected": -0.6796923279762268,
6452
+ "step": 842
6453
+ },
6454
+ {
6455
+ "epoch": 0.9774882374230909,
6456
+ "grad_norm": 82.61971695055652,
6457
+ "learning_rate": 1.1329202710118086e-07,
6458
+ "logits/chosen": -1.2287306785583496,
6459
+ "logits/rejected": -1.2248528003692627,
6460
+ "logps/chosen": -153.746826171875,
6461
+ "logps/rejected": -183.39047241210938,
6462
+ "loss": 0.5654,
6463
+ "rewards/accuracies": 0.625,
6464
+ "rewards/chosen": -0.13021187484264374,
6465
+ "rewards/margins": 0.843558132648468,
6466
+ "rewards/rejected": -0.9737700819969177,
6467
+ "step": 844
6468
+ },
6469
+ {
6470
+ "epoch": 0.9798045602605863,
6471
+ "grad_norm": 70.54919389801739,
6472
+ "learning_rate": 1.1290893786100058e-07,
6473
+ "logits/chosen": -1.2976385354995728,
6474
+ "logits/rejected": -1.2402337789535522,
6475
+ "logps/chosen": -132.8706817626953,
6476
+ "logps/rejected": -156.9254608154297,
6477
+ "loss": 0.5145,
6478
+ "rewards/accuracies": 0.78125,
6479
+ "rewards/chosen": -0.42142388224601746,
6480
+ "rewards/margins": 0.9857786297798157,
6481
+ "rewards/rejected": -1.4072024822235107,
6482
+ "step": 846
6483
+ },
6484
+ {
6485
+ "epoch": 0.9821208830980818,
6486
+ "grad_norm": 81.68100060471126,
6487
+ "learning_rate": 1.1252565586450131e-07,
6488
+ "logits/chosen": -1.2878555059432983,
6489
+ "logits/rejected": -1.337039828300476,
6490
+ "logps/chosen": -193.6406707763672,
6491
+ "logps/rejected": -220.8046112060547,
6492
+ "loss": 0.5546,
6493
+ "rewards/accuracies": 0.75,
6494
+ "rewards/chosen": -0.4571903347969055,
6495
+ "rewards/margins": 0.6591623425483704,
6496
+ "rewards/rejected": -1.1163526773452759,
6497
+ "step": 848
6498
+ },
6499
+ {
6500
+ "epoch": 0.9844372059355773,
6501
+ "grad_norm": 93.0455963225426,
6502
+ "learning_rate": 1.1214218683485157e-07,
6503
+ "logits/chosen": -1.3210810422897339,
6504
+ "logits/rejected": -1.2823312282562256,
6505
+ "logps/chosen": -187.9322509765625,
6506
+ "logps/rejected": -193.3695068359375,
6507
+ "loss": 0.5531,
6508
+ "rewards/accuracies": 0.71875,
6509
+ "rewards/chosen": -0.4511820077896118,
6510
+ "rewards/margins": 0.7341474294662476,
6511
+ "rewards/rejected": -1.1853294372558594,
6512
+ "step": 850
6513
+ },
6514
+ {
6515
+ "epoch": 0.9867535287730728,
6516
+ "grad_norm": 99.45149290799081,
6517
+ "learning_rate": 1.1175853649801273e-07,
6518
+ "logits/chosen": -1.2975995540618896,
6519
+ "logits/rejected": -1.3011971712112427,
6520
+ "logps/chosen": -134.58206176757812,
6521
+ "logps/rejected": -155.63638305664062,
6522
+ "loss": 0.6041,
6523
+ "rewards/accuracies": 0.65625,
6524
+ "rewards/chosen": -0.1595260500907898,
6525
+ "rewards/margins": 0.8379107713699341,
6526
+ "rewards/rejected": -0.9974368810653687,
6527
+ "step": 852
6528
+ },
6529
+ {
6530
+ "epoch": 0.9890698516105683,
6531
+ "grad_norm": 103.0078434001977,
6532
+ "learning_rate": 1.1137471058265329e-07,
6533
+ "logits/chosen": -1.2952150106430054,
6534
+ "logits/rejected": -1.3017812967300415,
6535
+ "logps/chosen": -148.48033142089844,
6536
+ "logps/rejected": -152.46951293945312,
6537
+ "loss": 0.5662,
6538
+ "rewards/accuracies": 0.59375,
6539
+ "rewards/chosen": -0.3374045491218567,
6540
+ "rewards/margins": 0.7678672671318054,
6541
+ "rewards/rejected": -1.105271816253662,
6542
+ "step": 854
6543
+ },
6544
+ {
6545
+ "epoch": 0.9913861744480637,
6546
+ "grad_norm": 132.30637646730733,
6547
+ "learning_rate": 1.1099071482006359e-07,
6548
+ "logits/chosen": -1.3640661239624023,
6549
+ "logits/rejected": -1.4117590188980103,
6550
+ "logps/chosen": -170.84063720703125,
6551
+ "logps/rejected": -178.2880096435547,
6552
+ "loss": 0.6467,
6553
+ "rewards/accuracies": 0.46875,
6554
+ "rewards/chosen": -0.8598957061767578,
6555
+ "rewards/margins": 0.3869331181049347,
6556
+ "rewards/rejected": -1.2468287944793701,
6557
+ "step": 856
6558
+ },
6559
+ {
6560
+ "epoch": 0.9937024972855591,
6561
+ "grad_norm": 106.55866485365634,
6562
+ "learning_rate": 1.1060655494407012e-07,
6563
+ "logits/chosen": -1.2328029870986938,
6564
+ "logits/rejected": -1.2614901065826416,
6565
+ "logps/chosen": -125.70144653320312,
6566
+ "logps/rejected": -148.56796264648438,
6567
+ "loss": 0.6174,
6568
+ "rewards/accuracies": 0.5625,
6569
+ "rewards/chosen": -0.24408888816833496,
6570
+ "rewards/margins": 0.35338127613067627,
6571
+ "rewards/rejected": -0.5974701642990112,
6572
+ "step": 858
6573
+ },
6574
+ {
6575
+ "epoch": 0.9960188201230546,
6576
+ "grad_norm": 92.00339838498802,
6577
+ "learning_rate": 1.102222366909499e-07,
6578
+ "logits/chosen": -1.125497579574585,
6579
+ "logits/rejected": -1.1907844543457031,
6580
+ "logps/chosen": -168.60272216796875,
6581
+ "logps/rejected": -184.2926025390625,
6582
+ "loss": 0.5696,
6583
+ "rewards/accuracies": 0.71875,
6584
+ "rewards/chosen": -0.504738450050354,
6585
+ "rewards/margins": 0.6949242353439331,
6586
+ "rewards/rejected": -1.199662685394287,
6587
+ "step": 860
6588
+ },
6589
+ {
6590
+ "epoch": 0.9983351429605501,
6591
+ "grad_norm": 95.4684250342951,
6592
+ "learning_rate": 1.0983776579934481e-07,
6593
+ "logits/chosen": -1.2691606283187866,
6594
+ "logits/rejected": -1.2375824451446533,
6595
+ "logps/chosen": -130.89102172851562,
6596
+ "logps/rejected": -144.6857452392578,
6597
+ "loss": 0.5968,
6598
+ "rewards/accuracies": 0.75,
6599
+ "rewards/chosen": -0.2888919413089752,
6600
+ "rewards/margins": 1.0368549823760986,
6601
+ "rewards/rejected": -1.3257468938827515,
6602
+ "step": 862
6603
+ },
6604
+ {
6605
+ "epoch": 1.0006514657980456,
6606
+ "grad_norm": 74.43449432333051,
6607
+ "learning_rate": 1.09453148010176e-07,
6608
+ "logits/chosen": -1.3525639772415161,
6609
+ "logits/rejected": -1.3091143369674683,
6610
+ "logps/chosen": -155.21115112304688,
6611
+ "logps/rejected": -155.0146942138672,
6612
+ "loss": 0.5571,
6613
+ "rewards/accuracies": 0.6875,
6614
+ "rewards/chosen": -0.15043863654136658,
6615
+ "rewards/margins": 0.7910802364349365,
6616
+ "rewards/rejected": -0.9415189027786255,
6617
+ "step": 864
6618
+ },
6619
+ {
6620
+ "epoch": 1.002967788635541,
6621
+ "grad_norm": 53.052072782970285,
6622
+ "learning_rate": 1.09068389066558e-07,
6623
+ "logits/chosen": -1.264554738998413,
6624
+ "logits/rejected": -1.3304630517959595,
6625
+ "logps/chosen": -138.06777954101562,
6626
+ "logps/rejected": -179.81643676757812,
6627
+ "loss": 0.4282,
6628
+ "rewards/accuracies": 0.75,
6629
+ "rewards/chosen": 0.0020294710993766785,
6630
+ "rewards/margins": 0.9756675362586975,
6631
+ "rewards/rejected": -0.9736379981040955,
6632
+ "step": 866
6633
+ },
6634
+ {
6635
+ "epoch": 1.0052841114730366,
6636
+ "grad_norm": 54.29727997773151,
6637
+ "learning_rate": 1.0868349471371314e-07,
6638
+ "logits/chosen": -1.2127741575241089,
6639
+ "logits/rejected": -1.2704501152038574,
6640
+ "logps/chosen": -142.79518127441406,
6641
+ "logps/rejected": -189.8652801513672,
6642
+ "loss": 0.4317,
6643
+ "rewards/accuracies": 0.78125,
6644
+ "rewards/chosen": -0.13422907888889313,
6645
+ "rewards/margins": 1.5447545051574707,
6646
+ "rewards/rejected": -1.678983449935913,
6647
+ "step": 868
6648
+ },
6649
+ {
6650
+ "epoch": 1.007600434310532,
6651
+ "grad_norm": 56.12931427605149,
6652
+ "learning_rate": 1.0829847069888565e-07,
6653
+ "logits/chosen": -1.2043306827545166,
6654
+ "logits/rejected": -1.2564079761505127,
6655
+ "logps/chosen": -111.9096908569336,
6656
+ "logps/rejected": -146.1145782470703,
6657
+ "loss": 0.463,
6658
+ "rewards/accuracies": 0.84375,
6659
+ "rewards/chosen": -0.0683784931898117,
6660
+ "rewards/margins": 0.8024751543998718,
6661
+ "rewards/rejected": -0.8708536624908447,
6662
+ "step": 870
6663
+ },
6664
+ {
6665
+ "epoch": 1.0099167571480274,
6666
+ "grad_norm": 51.89999261447005,
6667
+ "learning_rate": 1.0791332277125587e-07,
6668
+ "logits/chosen": -1.3936680555343628,
6669
+ "logits/rejected": -1.435255527496338,
6670
+ "logps/chosen": -146.84841918945312,
6671
+ "logps/rejected": -182.092041015625,
6672
+ "loss": 0.4189,
6673
+ "rewards/accuracies": 0.78125,
6674
+ "rewards/chosen": 0.1668233424425125,
6675
+ "rewards/margins": 1.2165982723236084,
6676
+ "rewards/rejected": -1.0497750043869019,
6677
+ "step": 872
6678
+ },
6679
+ {
6680
+ "epoch": 1.012233079985523,
6681
+ "grad_norm": 59.36325847754287,
6682
+ "learning_rate": 1.075280566818544e-07,
6683
+ "logits/chosen": -1.2359968423843384,
6684
+ "logits/rejected": -1.1934764385223389,
6685
+ "logps/chosen": -142.46665954589844,
6686
+ "logps/rejected": -150.53485107421875,
6687
+ "loss": 0.4029,
6688
+ "rewards/accuracies": 0.75,
6689
+ "rewards/chosen": 0.20680281519889832,
6690
+ "rewards/margins": 0.8232614994049072,
6691
+ "rewards/rejected": -0.6164587736129761,
6692
+ "step": 874
6693
+ },
6694
+ {
6695
+ "epoch": 1.0145494028230184,
6696
+ "grad_norm": 59.925719498657855,
6697
+ "learning_rate": 1.0714267818347629e-07,
6698
+ "logits/chosen": -1.351982831954956,
6699
+ "logits/rejected": -1.404510259628296,
6700
+ "logps/chosen": -119.25331115722656,
6701
+ "logps/rejected": -147.39370727539062,
6702
+ "loss": 0.4487,
6703
+ "rewards/accuracies": 0.84375,
6704
+ "rewards/chosen": -0.004702276550233364,
6705
+ "rewards/margins": 1.059253454208374,
6706
+ "rewards/rejected": -1.0639557838439941,
6707
+ "step": 876
6708
+ },
6709
+ {
6710
+ "epoch": 1.016865725660514,
6711
+ "grad_norm": 63.18524537189573,
6712
+ "learning_rate": 1.0675719303059492e-07,
6713
+ "logits/chosen": -1.3256399631500244,
6714
+ "logits/rejected": -1.351796269416809,
6715
+ "logps/chosen": -153.5157928466797,
6716
+ "logps/rejected": -202.40603637695312,
6717
+ "loss": 0.4485,
6718
+ "rewards/accuracies": 0.78125,
6719
+ "rewards/chosen": 0.10010495781898499,
6720
+ "rewards/margins": 1.2945441007614136,
6721
+ "rewards/rejected": -1.1944392919540405,
6722
+ "step": 878
6723
+ },
6724
+ {
6725
+ "epoch": 1.0191820484980094,
6726
+ "grad_norm": 56.63348082432867,
6727
+ "learning_rate": 1.0637160697927649e-07,
6728
+ "logits/chosen": -1.359007716178894,
6729
+ "logits/rejected": -1.3727188110351562,
6730
+ "logps/chosen": -164.15415954589844,
6731
+ "logps/rejected": -177.09884643554688,
6732
+ "loss": 0.4274,
6733
+ "rewards/accuracies": 0.8125,
6734
+ "rewards/chosen": -0.14782752096652985,
6735
+ "rewards/margins": 1.1597306728363037,
6736
+ "rewards/rejected": -1.307558298110962,
6737
+ "step": 880
6738
+ },
6739
+ {
6740
+ "epoch": 1.0214983713355048,
6741
+ "grad_norm": 47.49406015149724,
6742
+ "learning_rate": 1.059859257870936e-07,
6743
+ "logits/chosen": -1.2314316034317017,
6744
+ "logits/rejected": -1.3156919479370117,
6745
+ "logps/chosen": -99.96794891357422,
6746
+ "logps/rejected": -125.84103393554688,
6747
+ "loss": 0.4242,
6748
+ "rewards/accuracies": 0.78125,
6749
+ "rewards/chosen": 0.19646473228931427,
6750
+ "rewards/margins": 0.9843569993972778,
6751
+ "rewards/rejected": -0.78789222240448,
6752
+ "step": 882
6753
+ },
6754
+ {
6755
+ "epoch": 1.0238146941730004,
6756
+ "grad_norm": 45.83894890851922,
6757
+ "learning_rate": 1.0560015521303953e-07,
6758
+ "logits/chosen": -1.3569869995117188,
6759
+ "logits/rejected": -1.397745132446289,
6760
+ "logps/chosen": -140.70188903808594,
6761
+ "logps/rejected": -169.63076782226562,
6762
+ "loss": 0.3978,
6763
+ "rewards/accuracies": 0.8125,
6764
+ "rewards/chosen": 0.11355403065681458,
6765
+ "rewards/margins": 1.1430623531341553,
6766
+ "rewards/rejected": -1.029508352279663,
6767
+ "step": 884
6768
+ },
6769
+ {
6770
+ "epoch": 1.0261310170104958,
6771
+ "grad_norm": 54.040039551724966,
6772
+ "learning_rate": 1.0521430101744238e-07,
6773
+ "logits/chosen": -1.4817092418670654,
6774
+ "logits/rejected": -1.434765100479126,
6775
+ "logps/chosen": -151.61231994628906,
6776
+ "logps/rejected": -144.6493682861328,
6777
+ "loss": 0.4342,
6778
+ "rewards/accuracies": 0.6875,
6779
+ "rewards/chosen": 0.10845161974430084,
6780
+ "rewards/margins": 0.5633866786956787,
6781
+ "rewards/rejected": -0.4549350440502167,
6782
+ "step": 886
6783
+ },
6784
+ {
6785
+ "epoch": 1.0284473398479914,
6786
+ "grad_norm": 51.70194540758909,
6787
+ "learning_rate": 1.0482836896187861e-07,
6788
+ "logits/chosen": -1.207306981086731,
6789
+ "logits/rejected": -1.2579807043075562,
6790
+ "logps/chosen": -122.11793518066406,
6791
+ "logps/rejected": -145.04310607910156,
6792
+ "loss": 0.4389,
6793
+ "rewards/accuracies": 0.65625,
6794
+ "rewards/chosen": -0.15163260698318481,
6795
+ "rewards/margins": 0.5189149379730225,
6796
+ "rewards/rejected": -0.6705474853515625,
6797
+ "step": 888
6798
+ },
6799
+ {
6800
+ "epoch": 1.0307636626854868,
6801
+ "grad_norm": 61.2930931985187,
6802
+ "learning_rate": 1.044423648090875e-07,
6803
+ "logits/chosen": -1.3574914932250977,
6804
+ "logits/rejected": -1.445235013961792,
6805
+ "logps/chosen": -143.323974609375,
6806
+ "logps/rejected": -170.43026733398438,
6807
+ "loss": 0.4619,
6808
+ "rewards/accuracies": 0.84375,
6809
+ "rewards/chosen": 0.22330205142498016,
6810
+ "rewards/margins": 1.1949002742767334,
6811
+ "rewards/rejected": -0.971598207950592,
6812
+ "step": 890
6813
+ },
6814
+ {
6815
+ "epoch": 1.0330799855229822,
6816
+ "grad_norm": 52.83359389131165,
6817
+ "learning_rate": 1.0405629432288488e-07,
6818
+ "logits/chosen": -1.2581863403320312,
6819
+ "logits/rejected": -1.2796192169189453,
6820
+ "logps/chosen": -130.67315673828125,
6821
+ "logps/rejected": -158.39817810058594,
6822
+ "loss": 0.4437,
6823
+ "rewards/accuracies": 0.8125,
6824
+ "rewards/chosen": 2.531241625547409e-05,
6825
+ "rewards/margins": 1.0420689582824707,
6826
+ "rewards/rejected": -1.0420435667037964,
6827
+ "step": 892
6828
+ },
6829
+ {
6830
+ "epoch": 1.0353963083604778,
6831
+ "grad_norm": 54.354875880802155,
6832
+ "learning_rate": 1.036701632680769e-07,
6833
+ "logits/chosen": -1.2646857500076294,
6834
+ "logits/rejected": -1.2912808656692505,
6835
+ "logps/chosen": -143.00558471679688,
6836
+ "logps/rejected": -181.23011779785156,
6837
+ "loss": 0.4284,
6838
+ "rewards/accuracies": 0.90625,
6839
+ "rewards/chosen": -0.12129174172878265,
6840
+ "rewards/margins": 1.4072792530059814,
6841
+ "rewards/rejected": -1.5285711288452148,
6842
+ "step": 894
6843
+ },
6844
+ {
6845
+ "epoch": 1.0377126311979732,
6846
+ "grad_norm": 103.14604966973508,
6847
+ "learning_rate": 1.0328397741037426e-07,
6848
+ "logits/chosen": -1.3190052509307861,
6849
+ "logits/rejected": -1.2969970703125,
6850
+ "logps/chosen": -107.70980072021484,
6851
+ "logps/rejected": -113.95564270019531,
6852
+ "loss": 0.5385,
6853
+ "rewards/accuracies": 0.75,
6854
+ "rewards/chosen": -0.04918019846081734,
6855
+ "rewards/margins": 0.5177782773971558,
6856
+ "rewards/rejected": -0.566958487033844,
6857
+ "step": 896
6858
+ },
6859
+ {
6860
+ "epoch": 1.0400289540354688,
6861
+ "grad_norm": 43.79052026196426,
6862
+ "learning_rate": 1.0289774251630601e-07,
6863
+ "logits/chosen": -1.3973785638809204,
6864
+ "logits/rejected": -1.444900393486023,
6865
+ "logps/chosen": -96.34955596923828,
6866
+ "logps/rejected": -113.84618377685547,
6867
+ "loss": 0.3983,
6868
+ "rewards/accuracies": 0.78125,
6869
+ "rewards/chosen": 0.06471097469329834,
6870
+ "rewards/margins": 0.8261978626251221,
6871
+ "rewards/rejected": -0.7614869475364685,
6872
+ "step": 898
6873
+ },
6874
+ {
6875
+ "epoch": 1.0423452768729642,
6876
+ "grad_norm": 49.65418562930932,
6877
+ "learning_rate": 1.0251146435313328e-07,
6878
+ "logits/chosen": -1.3165615797042847,
6879
+ "logits/rejected": -1.294682502746582,
6880
+ "logps/chosen": -149.90878295898438,
6881
+ "logps/rejected": -151.5213623046875,
6882
+ "loss": 0.4286,
6883
+ "rewards/accuracies": 0.8125,
6884
+ "rewards/chosen": -0.07793111354112625,
6885
+ "rewards/margins": 0.9900121092796326,
6886
+ "rewards/rejected": -1.0679432153701782,
6887
+ "step": 900
6888
+ },
6889
+ {
6890
+ "epoch": 1.0423452768729642,
6891
+ "eval_logits/chosen": -1.267106533050537,
6892
+ "eval_logits/rejected": -1.2631834745407104,
6893
+ "eval_logps/chosen": -143.7148895263672,
6894
+ "eval_logps/rejected": -145.94981384277344,
6895
+ "eval_loss": 0.6367580890655518,
6896
+ "eval_rewards/accuracies": 0.6800000071525574,
6897
+ "eval_rewards/chosen": -0.8083698153495789,
6898
+ "eval_rewards/margins": 0.3912562429904938,
6899
+ "eval_rewards/rejected": -1.1996259689331055,
6900
+ "eval_runtime": 28.4118,
6901
+ "eval_samples_per_second": 3.52,
6902
+ "eval_steps_per_second": 0.88,
6903
+ "step": 900
6904
  }
6905
  ],
6906
  "logging_steps": 2,