Training in progress, epoch 3, checkpoint

last-checkpoint/global_step939/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e4a74e7eed9b560b26f85512d390d8088b3c07983ba6942e7f7e9287776a9ac
+size 13476836524

last-checkpoint/global_step939/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:fb5f3a94b2d368334faae0cfdbe0427d959986f3e11ca3079418c196fd359ef1
+size 13476836524

last-checkpoint/global_step939/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ffb4a63b8b4baee86c08604ed780a2d37b9f6a4d936ce98c28bd772a3cd6bc4
+size 13476836524

last-checkpoint/global_step939/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e24159d14871eaa97ee9f67ae2260c4b85bf832ef4c02f2a9fbee0a80e69375
+size 13476836524

last-checkpoint/global_step939/zero_pp_rank_0_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:253d93a65ababec7e8d2c5035bfcc45ae7a41810abb52fbdf0a05b83ccf04218
+size 150693

last-checkpoint/global_step939/zero_pp_rank_1_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ef94076c38828657a277c16da8f1894d0af86c6ec14a1796976e6f0c89f4315
+size 150693

last-checkpoint/global_step939/zero_pp_rank_2_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5565fd91b803663c46642d38c5588350a5093428d94b80b354ef4b8f8460d350
+size 150693

last-checkpoint/global_step939/zero_pp_rank_3_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10f60571c891579c183f2b7c26c677405ff7aa2b4558b2e2c86ad07d80334e6f
+size 150693

last-checkpoint/latest

@@ -1 +1 @@
-global_step626
+global_step939

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

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:37d98d0d79a2a8056e888825dbb89319bd37cc90e78333311f476731c0de96dd
+oid sha256:917f5b081e9bfa877d82ed3edf1360edb3e7b5f8925bd8218c89e3519c220613
 size 4938985352

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

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:98a0d6dc670bcc211a513dffddee9dab1866d26bfe1ddbab630b87c3d20efcb6
+oid sha256:087edec28f05c60443b011c4bf047332ddb25351d9602e3e589b59a3a8801ca0
 size 4947390880

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

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ac732abca0dd08baaa3ac8e6c052d653eb7d4789e9b99246e8456185f80cbf57
+oid sha256:dc5fe7a7275d5ec74ccef2f8351a8193871c676629eafdac6da2e5708c0f1a40
 size 3590488816

last-checkpoint/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2307c03867cef25b5028feb9a23f80e784b9af9a615de13ddca560a6a90fb593
+oid sha256:ab7cf0d34d60dfda516cf9661904550e2e294e723edd07c25c738f05e8ba92d1
 size 15024

last-checkpoint/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:50218cfaecdd818354e567b7167c13899e3b42297e7d8f58bd7e732cfa547800
+oid sha256:f7b183324e8227a51a9556d86b2ad893a8c4c52205ed4a737356c6611dac7353
 size 15024

last-checkpoint/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9814a66b49861f5495b06dae3be12ddf7185b88e2cae1fb808ca9efd99d5807f
+oid sha256:ac69e994090f4818cb1fa6f6cefa363178552c3c731c6507ff195bcb07fd5bef
 size 15024

last-checkpoint/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a7be93040a65e0a29975f6c70b94418e1fdf88423a50c58aa572141d3c92fbfc
+oid sha256:f68e61b63402f8afb1f69c960f7944965655dac11e3ccf29919c282f23931f86
 size 15024

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f7725b44c69725321786f1e58dccd7ea4d3ae5794ea47bd49c0d4a139aec266d
+oid sha256:fe7d17fbcfa5bee1bf6cb5aedab3ff5a70436912c200d7301d173be443809d63
 size 1064

last-checkpoint/trainer_state.json

@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.0,
+  "epoch": 3.0,
   "eval_steps": 500,
-  "global_step": 626,
+  "global_step": 939,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -984,6 +984,487 @@
       "eval_samples_per_second": 9.961,
       "eval_steps_per_second": 0.349,
       "step": 626
+    },
+    {
+      "epoch": 2.012779552715655,
+      "grad_norm": 0.003092817828863736,
+      "learning_rate": 1.8284023668639053e-07,
+      "logits/chosen": -0.7421875,
+      "logits/rejected": -0.4296875,
+      "logps/chosen": -316.0,
+      "logps/rejected": -262.0,
+      "loss": 0.0002,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.3515625,
+      "rewards/margins": 13.3125,
+      "rewards/rejected": -14.6875,
+      "step": 630
+    },
+    {
+      "epoch": 2.0447284345047922,
+      "grad_norm": 0.02088414959200188,
+      "learning_rate": 1.7692307692307693e-07,
+      "logits/chosen": -0.70703125,
+      "logits/rejected": -0.625,
+      "logps/chosen": -354.0,
+      "logps/rejected": -270.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.296875,
+      "rewards/margins": 13.3125,
+      "rewards/rejected": -14.5625,
+      "step": 640
+    },
+    {
+      "epoch": 2.07667731629393,
+      "grad_norm": 0.02978878745088855,
+      "learning_rate": 1.710059171597633e-07,
+      "logits/chosen": -0.5625,
+      "logits/rejected": -0.515625,
+      "logps/chosen": -326.0,
+      "logps/rejected": -260.0,
+      "loss": 0.0002,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.453125,
+      "rewards/margins": 13.375,
+      "rewards/rejected": -14.8125,
+      "step": 650
+    },
+    {
+      "epoch": 2.108626198083067,
+      "grad_norm": 0.014281695666503702,
+      "learning_rate": 1.650887573964497e-07,
+      "logits/chosen": -0.6640625,
+      "logits/rejected": -0.60546875,
+      "logps/chosen": -378.0,
+      "logps/rejected": -268.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.1875,
+      "rewards/margins": 13.625,
+      "rewards/rejected": -14.75,
+      "step": 660
+    },
+    {
+      "epoch": 2.1405750798722045,
+      "grad_norm": 0.1075200266687742,
+      "learning_rate": 1.591715976331361e-07,
+      "logits/chosen": -0.734375,
+      "logits/rejected": -0.447265625,
+      "logps/chosen": -306.0,
+      "logps/rejected": -266.0,
+      "loss": 0.0001,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.3515625,
+      "rewards/margins": 13.6875,
+      "rewards/rejected": -15.0625,
+      "step": 670
+    },
+    {
+      "epoch": 2.1725239616613417,
+      "grad_norm": 0.013374753124328204,
+      "learning_rate": 1.5325443786982248e-07,
+      "logits/chosen": -0.81640625,
+      "logits/rejected": -0.5234375,
+      "logps/chosen": -312.0,
+      "logps/rejected": -266.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.484375,
+      "rewards/margins": 13.5625,
+      "rewards/rejected": -15.0625,
+      "step": 680
+    },
+    {
+      "epoch": 2.2044728434504792,
+      "grad_norm": 0.05453915761092606,
+      "learning_rate": 1.4733727810650885e-07,
+      "logits/chosen": -0.6796875,
+      "logits/rejected": -0.50390625,
+      "logps/chosen": -298.0,
+      "logps/rejected": -262.0,
+      "loss": 0.0001,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.515625,
+      "rewards/margins": 13.25,
+      "rewards/rejected": -14.75,
+      "step": 690
+    },
+    {
+      "epoch": 2.236421725239617,
+      "grad_norm": 0.004507719187765237,
+      "learning_rate": 1.4142011834319526e-07,
+      "logits/chosen": -0.65625,
+      "logits/rejected": -0.55078125,
+      "logps/chosen": -302.0,
+      "logps/rejected": -276.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.5859375,
+      "rewards/margins": 13.625,
+      "rewards/rejected": -15.25,
+      "step": 700
+    },
+    {
+      "epoch": 2.268370607028754,
+      "grad_norm": 0.0014351004311762242,
+      "learning_rate": 1.3550295857988164e-07,
+      "logits/chosen": -0.78125,
+      "logits/rejected": -0.625,
+      "logps/chosen": -348.0,
+      "logps/rejected": -278.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.3828125,
+      "rewards/margins": 14.3125,
+      "rewards/rejected": -15.6875,
+      "step": 710
+    },
+    {
+      "epoch": 2.3003194888178915,
+      "grad_norm": 0.007661772736529076,
+      "learning_rate": 1.2958579881656802e-07,
+      "logits/chosen": -0.6328125,
+      "logits/rejected": -0.3984375,
+      "logps/chosen": -308.0,
+      "logps/rejected": -270.0,
+      "loss": 0.0001,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.25,
+      "rewards/margins": 14.0,
+      "rewards/rejected": -15.25,
+      "step": 720
+    },
+    {
+      "epoch": 2.3322683706070286,
+      "grad_norm": 0.09440998564444865,
+      "learning_rate": 1.2366863905325443e-07,
+      "logits/chosen": -0.60546875,
+      "logits/rejected": -0.51171875,
+      "logps/chosen": -380.0,
+      "logps/rejected": -272.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.1328125,
+      "rewards/margins": 14.3125,
+      "rewards/rejected": -15.4375,
+      "step": 730
+    },
+    {
+      "epoch": 2.364217252396166,
+      "grad_norm": 0.001050873951037513,
+      "learning_rate": 1.1775147928994082e-07,
+      "logits/chosen": -0.7109375,
+      "logits/rejected": -0.6484375,
+      "logps/chosen": -336.0,
+      "logps/rejected": -274.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.3046875,
+      "rewards/margins": 13.875,
+      "rewards/rejected": -15.1875,
+      "step": 740
+    },
+    {
+      "epoch": 2.3961661341853033,
+      "grad_norm": 0.0020244185172190623,
+      "learning_rate": 1.1183431952662721e-07,
+      "logits/chosen": -0.53515625,
+      "logits/rejected": -0.439453125,
+      "logps/chosen": -282.0,
+      "logps/rejected": -272.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.3359375,
+      "rewards/margins": 14.1875,
+      "rewards/rejected": -15.5625,
+      "step": 750
+    },
+    {
+      "epoch": 2.428115015974441,
+      "grad_norm": 0.0019816508240532195,
+      "learning_rate": 1.059171597633136e-07,
+      "logits/chosen": -0.66796875,
+      "logits/rejected": -0.515625,
+      "logps/chosen": -326.0,
+      "logps/rejected": -268.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.4609375,
+      "rewards/margins": 14.0,
+      "rewards/rejected": -15.4375,
+      "step": 760
+    },
+    {
+      "epoch": 2.460063897763578,
+      "grad_norm": 0.00974446806676642,
+      "learning_rate": 1e-07,
+      "logits/chosen": -0.6875,
+      "logits/rejected": -0.4921875,
+      "logps/chosen": -336.0,
+      "logps/rejected": -272.0,
+      "loss": 0.0002,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.453125,
+      "rewards/margins": 14.0,
+      "rewards/rejected": -15.5,
+      "step": 770
+    },
+    {
+      "epoch": 2.4920127795527156,
+      "grad_norm": 0.0013014264693432427,
+      "learning_rate": 9.408284023668639e-08,
+      "logits/chosen": -0.6796875,
+      "logits/rejected": -0.54296875,
+      "logps/chosen": -324.0,
+      "logps/rejected": -266.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.640625,
+      "rewards/margins": 13.5625,
+      "rewards/rejected": -15.1875,
+      "step": 780
+    },
+    {
+      "epoch": 2.523961661341853,
+      "grad_norm": 0.002562371187665821,
+      "learning_rate": 8.816568047337278e-08,
+      "logits/chosen": -0.67578125,
+      "logits/rejected": -0.5078125,
+      "logps/chosen": -354.0,
+      "logps/rejected": -268.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.40625,
+      "rewards/margins": 13.6875,
+      "rewards/rejected": -15.125,
+      "step": 790
+    },
+    {
+      "epoch": 2.5559105431309903,
+      "grad_norm": 0.000644996434656903,
+      "learning_rate": 8.224852071005916e-08,
+      "logits/chosen": -0.7265625,
+      "logits/rejected": -0.51171875,
+      "logps/chosen": -318.0,
+      "logps/rejected": -280.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.609375,
+      "rewards/margins": 14.5,
+      "rewards/rejected": -16.125,
+      "step": 800
+    },
+    {
+      "epoch": 2.587859424920128,
+      "grad_norm": 0.0008992028582889509,
+      "learning_rate": 7.633136094674555e-08,
+      "logits/chosen": -0.61328125,
+      "logits/rejected": -0.55078125,
+      "logps/chosen": -336.0,
+      "logps/rejected": -270.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.546875,
+      "rewards/margins": 14.375,
+      "rewards/rejected": -15.875,
+      "step": 810
+    },
+    {
+      "epoch": 2.619808306709265,
+      "grad_norm": 0.0012427097539449378,
+      "learning_rate": 7.041420118343195e-08,
+      "logits/chosen": -0.62890625,
+      "logits/rejected": -0.546875,
+      "logps/chosen": -354.0,
+      "logps/rejected": -278.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.328125,
+      "rewards/margins": 14.3125,
+      "rewards/rejected": -15.6875,
+      "step": 820
+    },
+    {
+      "epoch": 2.6517571884984026,
+      "grad_norm": 0.0020240111576199865,
+      "learning_rate": 6.449704142011835e-08,
+      "logits/chosen": -0.73828125,
+      "logits/rejected": -0.53515625,
+      "logps/chosen": -308.0,
+      "logps/rejected": -268.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.3203125,
+      "rewards/margins": 14.0,
+      "rewards/rejected": -15.375,
+      "step": 830
+    },
+    {
+      "epoch": 2.68370607028754,
+      "grad_norm": 0.001503960612923789,
+      "learning_rate": 5.857988165680473e-08,
+      "logits/chosen": -0.640625,
+      "logits/rejected": -0.5,
+      "logps/chosen": -288.0,
+      "logps/rejected": -276.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.2890625,
+      "rewards/margins": 14.375,
+      "rewards/rejected": -15.6875,
+      "step": 840
+    },
+    {
+      "epoch": 2.7156549520766773,
+      "grad_norm": 0.0010888686135269838,
+      "learning_rate": 5.266272189349112e-08,
+      "logits/chosen": -0.703125,
+      "logits/rejected": -0.546875,
+      "logps/chosen": -326.0,
+      "logps/rejected": -274.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.625,
+      "rewards/margins": 13.875,
+      "rewards/rejected": -15.5,
+      "step": 850
+    },
+    {
+      "epoch": 2.747603833865815,
+      "grad_norm": 0.003491886927346326,
+      "learning_rate": 4.674556213017751e-08,
+      "logits/chosen": -0.58984375,
+      "logits/rejected": -0.48046875,
+      "logps/chosen": -348.0,
+      "logps/rejected": -270.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.34375,
+      "rewards/margins": 14.125,
+      "rewards/rejected": -15.4375,
+      "step": 860
+    },
+    {
+      "epoch": 2.779552715654952,
+      "grad_norm": 0.0003618045449194657,
+      "learning_rate": 4.082840236686391e-08,
+      "logits/chosen": -0.74609375,
+      "logits/rejected": -0.53125,
+      "logps/chosen": -352.0,
+      "logps/rejected": -276.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.4609375,
+      "rewards/margins": 14.25,
+      "rewards/rejected": -15.6875,
+      "step": 870
+    },
+    {
+      "epoch": 2.8115015974440896,
+      "grad_norm": 0.0020187053364208985,
+      "learning_rate": 3.4911242603550294e-08,
+      "logits/chosen": -0.69921875,
+      "logits/rejected": -0.50390625,
+      "logps/chosen": -258.0,
+      "logps/rejected": -272.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.5078125,
+      "rewards/margins": 14.1875,
+      "rewards/rejected": -15.6875,
+      "step": 880
+    },
+    {
+      "epoch": 2.8434504792332267,
+      "grad_norm": 0.0010734719479623244,
+      "learning_rate": 2.8994082840236687e-08,
+      "logits/chosen": -0.578125,
+      "logits/rejected": -0.51171875,
+      "logps/chosen": -382.0,
+      "logps/rejected": -268.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.34375,
+      "rewards/margins": 13.8125,
+      "rewards/rejected": -15.125,
+      "step": 890
+    },
+    {
+      "epoch": 2.8753993610223643,
+      "grad_norm": 0.0007849148592248786,
+      "learning_rate": 2.3076923076923076e-08,
+      "logits/chosen": -0.734375,
+      "logits/rejected": -0.47265625,
+      "logps/chosen": -278.0,
+      "logps/rejected": -272.0,
+      "loss": 0.0001,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.6796875,
+      "rewards/margins": 14.1875,
+      "rewards/rejected": -15.875,
+      "step": 900
+    },
+    {
+      "epoch": 2.9073482428115014,
+      "grad_norm": 0.0009071422276518199,
+      "learning_rate": 1.7159763313609465e-08,
+      "logits/chosen": -0.546875,
+      "logits/rejected": -0.32421875,
+      "logps/chosen": -324.0,
+      "logps/rejected": -262.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.5859375,
+      "rewards/margins": 14.0625,
+      "rewards/rejected": -15.625,
+      "step": 910
+    },
+    {
+      "epoch": 2.939297124600639,
+      "grad_norm": 0.02336710919043686,
+      "learning_rate": 1.1242603550295858e-08,
+      "logits/chosen": -0.765625,
+      "logits/rejected": -0.427734375,
+      "logps/chosen": -348.0,
+      "logps/rejected": -268.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.390625,
+      "rewards/margins": 14.1875,
+      "rewards/rejected": -15.625,
+      "step": 920
+    },
+    {
+      "epoch": 2.9712460063897765,
+      "grad_norm": 0.03716794354422435,
+      "learning_rate": 5.325443786982248e-09,
+      "logits/chosen": -0.7421875,
+      "logits/rejected": -0.578125,
+      "logps/chosen": -326.0,
+      "logps/rejected": -272.0,
+      "loss": 0.0,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -1.515625,
+      "rewards/margins": 14.0625,
+      "rewards/rejected": -15.5625,
+      "step": 930
+    },
+    {
+      "epoch": 3.0,
+      "eval_logits/chosen": -0.71875,
+      "eval_logits/rejected": -0.44921875,
+      "eval_logps/chosen": -306.0,
+      "eval_logps/rejected": -268.0,
+      "eval_loss": 7.638990791747347e-05,
+      "eval_rewards/accuracies": 1.0,
+      "eval_rewards/chosen": -1.3515625,
+      "eval_rewards/margins": 13.9375,
+      "eval_rewards/rejected": -15.3125,
+      "eval_runtime": 20.8014,
+      "eval_samples_per_second": 9.615,
+      "eval_steps_per_second": 0.337,
+      "step": 939
     }
   ],
   "logging_steps": 10,
@@ -998,7 +1479,7 @@
         "should_evaluate": false,
         "should_log": false,
         "should_save": true,
-        "should_training_stop": false
+        "should_training_stop": true
       },
       "attributes": {}
     }