Training in progress, epoch 2, checkpoint

last-checkpoint/global_step626/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c127a426ee0ed2f8fd4d71b092602e7ee22b124351c2427398eedd6b76f2c3a
+size 14483467880

last-checkpoint/global_step626/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:917a99521ebd96803932d7f16a91a6a93b00ecf9857d0261d04f6f8e86f5e63a
+size 14483467880

last-checkpoint/global_step626/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c2cd99045ee5b27e8a3af203f91730fac31617536d565f851483be4e8129a9b
+size 14483467880

last-checkpoint/global_step626/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed033c05657b23b0cdb95291329a8abe66acbb63c39efb1b86aec926937aec14
+size 14483467880

last-checkpoint/global_step626/zero_pp_rank_0_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:187ffe92ce55235bbe82f73ea0b6543ca835e4fdeebc034963b25816f3ad3892
+size 150629

last-checkpoint/global_step626/zero_pp_rank_1_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ecc916b0f8a257c28cd0bb1cdccbcf7f0e8c32ef382b46074def8b60d3b2b492
+size 150629

last-checkpoint/global_step626/zero_pp_rank_2_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09bb50354b2e9b768e7f5879c4260ffe34cff8b5c6fff67fa591598195ef9284
+size 150629

last-checkpoint/global_step626/zero_pp_rank_3_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6ee386962043fb3f07eea85ab7ffa2b7f60c7e7ca6a648a3922335ea1bb147d
+size 150629

last-checkpoint/latest

@@ -1 +1 @@
-global_step313
+global_step626

last-checkpoint/model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c151d3016900a69e66cecf0f47240a8df6135383d1cc89185987cbe5815348ae
+oid sha256:f1ecb90f499c5b372ce8957768864282a6551fee2eb3bd7a51654fe59c8a46f0
 size 4943162336

last-checkpoint/model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:04e0ea297068a1763cea90c85ac8da2de5f5fb535db9cf0998b97b1f8b34d2f5
+oid sha256:f2fade6fae487a72b60a9e4251f69b61c1249016a230b660d27f2ddedab470bf
 size 4999819336

last-checkpoint/model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:01806d79474b0b2d611460746d886dbfe05133a01b21ed95d98fc0343acb9716
+oid sha256:bdf4b20bc902a231700c052ca140f23c4a223a60d352cdd14347e96514f78cff
 size 4540516344

last-checkpoint/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bbe0d720c4c75a6a04213fa3b64bacbe794718a53e2b56ebb67a1a795014dfad
+oid sha256:70cc56408014c410353d4dd58ae9b03f4be043f5f800324f66fd8e20e99b840e
 size 15024

last-checkpoint/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:72452d3138d0ca2ff89429e3294a834ae7a68e8596fc757735ca56ae52509d57
+oid sha256:49d1438e98cc9c53a6852464635ce62e9788e61eb3646b73e33813f487c4b6ae
 size 15024

last-checkpoint/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f36e306fb8ebcf53a167bfd6c9af74db410a269ada1e619e3e816f5269543b9d
+oid sha256:4388add9cec90932f8ff0100d27a0574d98e1bad52ff89d44e31967d2b4fbfde
 size 15024

last-checkpoint/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bb47ce0c6f815a6f8302b0e3819b4c2315ca71dae3138d97fdceb765cdd0a039
+oid sha256:a705d6dfaae4f2c1b4b2be6b25a6eb521ffae6fcba21cc1531e97b60037ed079
 size 15024

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fe577d7bcf08e961c71c0d5dde48713ba4caaef0df5933e3a48fcf4e7227dad4
+oid sha256:def86922cfbc44372ebbbb33f1d7fca89d594cb9a47912a7c87d6e97ad52cc09
 size 1064

last-checkpoint/trainer_state.json

@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.0,
+  "epoch": 2.0,
   "eval_steps": 500,
-  "global_step": 313,
+  "global_step": 626,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -503,6 +503,487 @@
       "eval_samples_per_second": 16.464,
       "eval_steps_per_second": 0.576,
       "step": 313
+    },
+    {
+      "epoch": 1.0223642172523961,
+      "grad_norm": 12.780435839397535,
+      "learning_rate": 3.662721893491124e-07,
+      "logits/chosen": -3.859375,
+      "logits/rejected": -3.859375,
+      "logps/chosen": -144.0,
+      "logps/rejected": -154.0,
+      "loss": 0.5925,
+      "rewards/accuracies": 0.25,
+      "rewards/chosen": -0.06494140625,
+      "rewards/margins": 0.84765625,
+      "rewards/rejected": -0.91015625,
+      "step": 320
+    },
+    {
+      "epoch": 1.0543130990415335,
+      "grad_norm": 2.3499904446036433,
+      "learning_rate": 3.603550295857988e-07,
+      "logits/chosen": -3.875,
+      "logits/rejected": -3.90625,
+      "logps/chosen": -129.0,
+      "logps/rejected": -146.0,
+      "loss": 0.54,
+      "rewards/accuracies": 0.21250000596046448,
+      "rewards/chosen": -0.50390625,
+      "rewards/margins": 0.6640625,
+      "rewards/rejected": -1.171875,
+      "step": 330
+    },
+    {
+      "epoch": 1.0862619808306708,
+      "grad_norm": 5.297851157977379,
+      "learning_rate": 3.544378698224852e-07,
+      "logits/chosen": -3.9375,
+      "logits/rejected": -3.984375,
+      "logps/chosen": -133.0,
+      "logps/rejected": -143.0,
+      "loss": 0.5106,
+      "rewards/accuracies": 0.30000001192092896,
+      "rewards/chosen": -0.2890625,
+      "rewards/margins": 0.8828125,
+      "rewards/rejected": -1.171875,
+      "step": 340
+    },
+    {
+      "epoch": 1.1182108626198084,
+      "grad_norm": 15.356893947406501,
+      "learning_rate": 3.485207100591716e-07,
+      "logits/chosen": -3.84375,
+      "logits/rejected": -3.84375,
+      "logps/chosen": -127.0,
+      "logps/rejected": -158.0,
+      "loss": 0.5021,
+      "rewards/accuracies": 0.2874999940395355,
+      "rewards/chosen": -0.41015625,
+      "rewards/margins": 1.3515625,
+      "rewards/rejected": -1.7578125,
+      "step": 350
+    },
+    {
+      "epoch": 1.1501597444089458,
+      "grad_norm": 7.611884421622937,
+      "learning_rate": 3.42603550295858e-07,
+      "logits/chosen": -3.875,
+      "logits/rejected": -3.96875,
+      "logps/chosen": -145.0,
+      "logps/rejected": -178.0,
+      "loss": 0.5444,
+      "rewards/accuracies": 0.22499999403953552,
+      "rewards/chosen": -1.25,
+      "rewards/margins": 1.109375,
+      "rewards/rejected": -2.359375,
+      "step": 360
+    },
+    {
+      "epoch": 1.182108626198083,
+      "grad_norm": 14.348371026442878,
+      "learning_rate": 3.366863905325444e-07,
+      "logits/chosen": -4.0,
+      "logits/rejected": -4.0625,
+      "logps/chosen": -158.0,
+      "logps/rejected": -170.0,
+      "loss": 0.5288,
+      "rewards/accuracies": 0.20000000298023224,
+      "rewards/chosen": -1.6640625,
+      "rewards/margins": 1.0625,
+      "rewards/rejected": -2.734375,
+      "step": 370
+    },
+    {
+      "epoch": 1.2140575079872205,
+      "grad_norm": 2.947572631593703,
+      "learning_rate": 3.3076923076923075e-07,
+      "logits/chosen": -3.84375,
+      "logits/rejected": -3.78125,
+      "logps/chosen": -129.0,
+      "logps/rejected": -141.0,
+      "loss": 0.5245,
+      "rewards/accuracies": 0.20000000298023224,
+      "rewards/chosen": -1.1171875,
+      "rewards/margins": 0.9453125,
+      "rewards/rejected": -2.0625,
+      "step": 380
+    },
+    {
+      "epoch": 1.2460063897763578,
+      "grad_norm": 8.284381008128047,
+      "learning_rate": 3.2485207100591716e-07,
+      "logits/chosen": -3.796875,
+      "logits/rejected": -3.84375,
+      "logps/chosen": -146.0,
+      "logps/rejected": -165.0,
+      "loss": 0.5533,
+      "rewards/accuracies": 0.23749999701976776,
+      "rewards/chosen": -1.265625,
+      "rewards/margins": 1.0078125,
+      "rewards/rejected": -2.265625,
+      "step": 390
+    },
+    {
+      "epoch": 1.2779552715654952,
+      "grad_norm": 3.2439367204723837,
+      "learning_rate": 3.189349112426035e-07,
+      "logits/chosen": -3.890625,
+      "logits/rejected": -3.84375,
+      "logps/chosen": -160.0,
+      "logps/rejected": -176.0,
+      "loss": 0.5292,
+      "rewards/accuracies": 0.30000001192092896,
+      "rewards/chosen": -1.671875,
+      "rewards/margins": 1.1328125,
+      "rewards/rejected": -2.796875,
+      "step": 400
+    },
+    {
+      "epoch": 1.3099041533546325,
+      "grad_norm": 23.291826626822854,
+      "learning_rate": 3.130177514792899e-07,
+      "logits/chosen": -3.9375,
+      "logits/rejected": -3.90625,
+      "logps/chosen": -149.0,
+      "logps/rejected": -167.0,
+      "loss": 0.4924,
+      "rewards/accuracies": 0.32499998807907104,
+      "rewards/chosen": -1.625,
+      "rewards/margins": 1.2578125,
+      "rewards/rejected": -2.875,
+      "step": 410
+    },
+    {
+      "epoch": 1.34185303514377,
+      "grad_norm": 11.526112822046265,
+      "learning_rate": 3.071005917159763e-07,
+      "logits/chosen": -3.828125,
+      "logits/rejected": -3.96875,
+      "logps/chosen": -148.0,
+      "logps/rejected": -175.0,
+      "loss": 0.4906,
+      "rewards/accuracies": 0.38749998807907104,
+      "rewards/chosen": -1.1328125,
+      "rewards/margins": 1.8671875,
+      "rewards/rejected": -3.0,
+      "step": 420
+    },
+    {
+      "epoch": 1.3738019169329074,
+      "grad_norm": 7.079418300270812,
+      "learning_rate": 3.011834319526627e-07,
+      "logits/chosen": -3.765625,
+      "logits/rejected": -3.84375,
+      "logps/chosen": -137.0,
+      "logps/rejected": -163.0,
+      "loss": 0.5138,
+      "rewards/accuracies": 0.2750000059604645,
+      "rewards/chosen": -1.625,
+      "rewards/margins": 1.1328125,
+      "rewards/rejected": -2.765625,
+      "step": 430
+    },
+    {
+      "epoch": 1.4057507987220448,
+      "grad_norm": 23.51606757056333,
+      "learning_rate": 2.952662721893491e-07,
+      "logits/chosen": -3.921875,
+      "logits/rejected": -3.890625,
+      "logps/chosen": -149.0,
+      "logps/rejected": -164.0,
+      "loss": 0.5057,
+      "rewards/accuracies": 0.23749999701976776,
+      "rewards/chosen": -1.890625,
+      "rewards/margins": 1.140625,
+      "rewards/rejected": -3.03125,
+      "step": 440
+    },
+    {
+      "epoch": 1.4376996805111821,
+      "grad_norm": 11.56064802099853,
+      "learning_rate": 2.893491124260355e-07,
+      "logits/chosen": -3.8125,
+      "logits/rejected": -3.8125,
+      "logps/chosen": -144.0,
+      "logps/rejected": -174.0,
+      "loss": 0.5141,
+      "rewards/accuracies": 0.32499998807907104,
+      "rewards/chosen": -1.4609375,
+      "rewards/margins": 1.6640625,
+      "rewards/rejected": -3.125,
+      "step": 450
+    },
+    {
+      "epoch": 1.4696485623003195,
+      "grad_norm": 11.236096530357507,
+      "learning_rate": 2.834319526627219e-07,
+      "logits/chosen": -3.75,
+      "logits/rejected": -3.75,
+      "logps/chosen": -152.0,
+      "logps/rejected": -165.0,
+      "loss": 0.5067,
+      "rewards/accuracies": 0.22499999403953552,
+      "rewards/chosen": -1.9609375,
+      "rewards/margins": 1.15625,
+      "rewards/rejected": -3.125,
+      "step": 460
+    },
+    {
+      "epoch": 1.5015974440894568,
+      "grad_norm": 3.370024966920903,
+      "learning_rate": 2.7751479289940824e-07,
+      "logits/chosen": -3.921875,
+      "logits/rejected": -3.890625,
+      "logps/chosen": -141.0,
+      "logps/rejected": -163.0,
+      "loss": 0.523,
+      "rewards/accuracies": 0.32499998807907104,
+      "rewards/chosen": -2.09375,
+      "rewards/margins": 1.53125,
+      "rewards/rejected": -3.625,
+      "step": 470
+    },
+    {
+      "epoch": 1.5335463258785942,
+      "grad_norm": 9.978824743164745,
+      "learning_rate": 2.715976331360947e-07,
+      "logits/chosen": -3.828125,
+      "logits/rejected": -3.875,
+      "logps/chosen": -145.0,
+      "logps/rejected": -163.0,
+      "loss": 0.5252,
+      "rewards/accuracies": 0.21250000596046448,
+      "rewards/chosen": -2.5,
+      "rewards/margins": 0.96484375,
+      "rewards/rejected": -3.453125,
+      "step": 480
+    },
+    {
+      "epoch": 1.5654952076677318,
+      "grad_norm": 14.326969919306727,
+      "learning_rate": 2.6568047337278106e-07,
+      "logits/chosen": -3.9375,
+      "logits/rejected": -3.984375,
+      "logps/chosen": -152.0,
+      "logps/rejected": -172.0,
+      "loss": 0.5329,
+      "rewards/accuracies": 0.25,
+      "rewards/chosen": -2.515625,
+      "rewards/margins": 1.734375,
+      "rewards/rejected": -4.25,
+      "step": 490
+    },
+    {
+      "epoch": 1.5974440894568689,
+      "grad_norm": 6.967846297592076,
+      "learning_rate": 2.5976331360946746e-07,
+      "logits/chosen": -4.03125,
+      "logits/rejected": -4.0625,
+      "logps/chosen": -176.0,
+      "logps/rejected": -205.0,
+      "loss": 0.539,
+      "rewards/accuracies": 0.2750000059604645,
+      "rewards/chosen": -2.75,
+      "rewards/margins": 1.53125,
+      "rewards/rejected": -4.28125,
+      "step": 500
+    },
+    {
+      "epoch": 1.6293929712460065,
+      "grad_norm": 10.478631685036758,
+      "learning_rate": 2.538461538461538e-07,
+      "logits/chosen": -3.96875,
+      "logits/rejected": -3.953125,
+      "logps/chosen": -166.0,
+      "logps/rejected": -181.0,
+      "loss": 0.5138,
+      "rewards/accuracies": 0.2874999940395355,
+      "rewards/chosen": -2.390625,
+      "rewards/margins": 1.53125,
+      "rewards/rejected": -3.921875,
+      "step": 510
+    },
+    {
+      "epoch": 1.6613418530351438,
+      "grad_norm": 3.9485916094762277,
+      "learning_rate": 2.479289940828402e-07,
+      "logits/chosen": -3.8125,
+      "logits/rejected": -3.859375,
+      "logps/chosen": -138.0,
+      "logps/rejected": -165.0,
+      "loss": 0.5187,
+      "rewards/accuracies": 0.3375000059604645,
+      "rewards/chosen": -1.78125,
+      "rewards/margins": 1.7890625,
+      "rewards/rejected": -3.578125,
+      "step": 520
+    },
+    {
+      "epoch": 1.6932907348242812,
+      "grad_norm": 9.691106682722046,
+      "learning_rate": 2.420118343195266e-07,
+      "logits/chosen": -3.953125,
+      "logits/rejected": -3.96875,
+      "logps/chosen": -149.0,
+      "logps/rejected": -163.0,
+      "loss": 0.5132,
+      "rewards/accuracies": 0.26249998807907104,
+      "rewards/chosen": -1.890625,
+      "rewards/margins": 1.0078125,
+      "rewards/rejected": -2.890625,
+      "step": 530
+    },
+    {
+      "epoch": 1.7252396166134185,
+      "grad_norm": 6.194603425697891,
+      "learning_rate": 2.36094674556213e-07,
+      "logits/chosen": -3.875,
+      "logits/rejected": -3.90625,
+      "logps/chosen": -158.0,
+      "logps/rejected": -181.0,
+      "loss": 0.5271,
+      "rewards/accuracies": 0.23749999701976776,
+      "rewards/chosen": -2.328125,
+      "rewards/margins": 1.2890625,
+      "rewards/rejected": -3.609375,
+      "step": 540
+    },
+    {
+      "epoch": 1.7571884984025559,
+      "grad_norm": 4.8547868881787535,
+      "learning_rate": 2.301775147928994e-07,
+      "logits/chosen": -3.875,
+      "logits/rejected": -3.890625,
+      "logps/chosen": -158.0,
+      "logps/rejected": -180.0,
+      "loss": 0.5112,
+      "rewards/accuracies": 0.2750000059604645,
+      "rewards/chosen": -2.265625,
+      "rewards/margins": 1.6015625,
+      "rewards/rejected": -3.875,
+      "step": 550
+    },
+    {
+      "epoch": 1.7891373801916934,
+      "grad_norm": 6.042788937636992,
+      "learning_rate": 2.242603550295858e-07,
+      "logits/chosen": -3.875,
+      "logits/rejected": -3.90625,
+      "logps/chosen": -157.0,
+      "logps/rejected": -203.0,
+      "loss": 0.5342,
+      "rewards/accuracies": 0.3125,
+      "rewards/chosen": -2.8125,
+      "rewards/margins": 1.65625,
+      "rewards/rejected": -4.46875,
+      "step": 560
+    },
+    {
+      "epoch": 1.8210862619808306,
+      "grad_norm": 17.608362227397397,
+      "learning_rate": 2.1834319526627217e-07,
+      "logits/chosen": -3.75,
+      "logits/rejected": -3.859375,
+      "logps/chosen": -140.0,
+      "logps/rejected": -173.0,
+      "loss": 0.5249,
+      "rewards/accuracies": 0.26249998807907104,
+      "rewards/chosen": -2.125,
+      "rewards/margins": 1.609375,
+      "rewards/rejected": -3.734375,
+      "step": 570
+    },
+    {
+      "epoch": 1.8530351437699681,
+      "grad_norm": 24.398412082273794,
+      "learning_rate": 2.1242603550295858e-07,
+      "logits/chosen": -3.78125,
+      "logits/rejected": -3.734375,
+      "logps/chosen": -151.0,
+      "logps/rejected": -158.0,
+      "loss": 0.52,
+      "rewards/accuracies": 0.25,
+      "rewards/chosen": -2.953125,
+      "rewards/margins": 1.171875,
+      "rewards/rejected": -4.125,
+      "step": 580
+    },
+    {
+      "epoch": 1.8849840255591053,
+      "grad_norm": 17.069584368649224,
+      "learning_rate": 2.0650887573964496e-07,
+      "logits/chosen": -3.796875,
+      "logits/rejected": -3.8125,
+      "logps/chosen": -165.0,
+      "logps/rejected": -181.0,
+      "loss": 0.5235,
+      "rewards/accuracies": 0.2750000059604645,
+      "rewards/chosen": -3.28125,
+      "rewards/margins": 1.5390625,
+      "rewards/rejected": -4.8125,
+      "step": 590
+    },
+    {
+      "epoch": 1.9169329073482428,
+      "grad_norm": 2.1180201542039883,
+      "learning_rate": 2.0059171597633133e-07,
+      "logits/chosen": -3.765625,
+      "logits/rejected": -3.84375,
+      "logps/chosen": -150.0,
+      "logps/rejected": -176.0,
+      "loss": 0.5272,
+      "rewards/accuracies": 0.2750000059604645,
+      "rewards/chosen": -3.15625,
+      "rewards/margins": 1.6015625,
+      "rewards/rejected": -4.75,
+      "step": 600
+    },
+    {
+      "epoch": 1.9488817891373802,
+      "grad_norm": 3.4777849704272694,
+      "learning_rate": 1.9467455621301774e-07,
+      "logits/chosen": -3.84375,
+      "logits/rejected": -3.84375,
+      "logps/chosen": -152.0,
+      "logps/rejected": -160.0,
+      "loss": 0.5547,
+      "rewards/accuracies": 0.26249998807907104,
+      "rewards/chosen": -2.6875,
+      "rewards/margins": 1.203125,
+      "rewards/rejected": -3.890625,
+      "step": 610
+    },
+    {
+      "epoch": 1.9808306709265175,
+      "grad_norm": 27.898326722256666,
+      "learning_rate": 1.8875739644970412e-07,
+      "logits/chosen": -3.5625,
+      "logits/rejected": -3.5625,
+      "logps/chosen": -160.0,
+      "logps/rejected": -174.0,
+      "loss": 0.5474,
+      "rewards/accuracies": 0.22499999403953552,
+      "rewards/chosen": -3.390625,
+      "rewards/margins": 1.1015625,
+      "rewards/rejected": -4.5,
+      "step": 620
+    },
+    {
+      "epoch": 2.0,
+      "eval_logits/chosen": -3.8125,
+      "eval_logits/rejected": -3.828125,
+      "eval_logps/chosen": -144.0,
+      "eval_logps/rejected": -151.0,
+      "eval_loss": 0.6862499713897705,
+      "eval_rewards/accuracies": 0.1428571492433548,
+      "eval_rewards/chosen": -3.4375,
+      "eval_rewards/margins": 0.33203125,
+      "eval_rewards/rejected": -3.78125,
+      "eval_runtime": 12.0804,
+      "eval_samples_per_second": 16.556,
+      "eval_steps_per_second": 0.579,
+      "step": 626
     }
   ],
   "logging_steps": 10,