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 sha256:8d9300ca7bbda237b4b9027720607655238a3d48a377a6c6daa70e68edc18fc0
+size 14496051368

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 sha256:9d003a9e0cc9423ca19378739933d308d6bd9f02908081ec89fc8be3415545d9
+size 14496051368

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 sha256:8bc2a7e471b555f913d0060f61bae1057b66cb0cbe13f0c368f528b2d20fc947
+size 14496051368

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 sha256:0cfaf2c7ba3543457cd28a2ae36272aefa0dc2d691f4a897c9a9dcf4f7cdf181
+size 14496051368

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:b2efe7d77b4410e6649a35dc6efabb73a0dfc5153beda8cefa2495c69bf65fbe
+size 150629

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 sha256:3e6cad11b1fd5787adcea9f18e4c482897157146658af7358b8e1330c5de61c5
+size 150629

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 sha256:6eb281386924df332f5956c9354cf760b10785f53e4738767a85453249c0b725
+size 150629

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 sha256:32f520b4a0c187f92c6bf6c96396edb2cd0df53e8944e38b0ce6f06db6b59b05
+size 150629

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:fcde0e3fd57d3565a5efc68083353109552e415e0e3ac697493bdd227fa8ea90
+oid sha256:2438d764a87b5af08c71585fc7e6bfa26fe838ea911a97f41ca0ce0833740481
 size 4949453792

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

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:83735bc0086fb3a08f3edfdf73435ec418f166a6d26d66d09ce6ff7de969afc3
+oid sha256:f7830842c04eb0b0f0f50b81978c5155cd89f2f98555f3230259666dd07eb3bb
 size 4999819336

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

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f61596553e3d699a5e29ac85de39be5889c2e07950b4e7a6d7bf942b8f3e1cff
+oid sha256:cdc08721565b56e6db4df3ee97941c45162528ee6d6269d018f4a32a40bb7f68
 size 4546807800

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:c3fd160e69e5766b513d913785299b2ff6d181cae9fd47f741adc100cb17d22e
+oid sha256:640747ca69b57ef58ae079eec14158d13652c8bf30fb7c733fab30d65edaeb50
 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": 16.441,
       "eval_steps_per_second": 0.575,
       "step": 626
+    },
+    {
+      "epoch": 2.012779552715655,
+      "grad_norm": 0.08332814736176879,
+      "learning_rate": 1.8284023668639053e-07,
+      "logits/chosen": -0.96875,
+      "logits/rejected": -1.1171875,
+      "logps/chosen": -74.5,
+      "logps/rejected": -83.0,
+      "loss": 0.6226,
+      "rewards/accuracies": 0.0625,
+      "rewards/chosen": -4.84375,
+      "rewards/margins": 0.671875,
+      "rewards/rejected": -5.53125,
+      "step": 630
+    },
+    {
+      "epoch": 2.0447284345047922,
+      "grad_norm": 0.164985938182988,
+      "learning_rate": 1.7692307692307693e-07,
+      "logits/chosen": -1.1796875,
+      "logits/rejected": -1.09375,
+      "logps/chosen": -67.5,
+      "logps/rejected": -82.0,
+      "loss": 0.6191,
+      "rewards/accuracies": 0.13750000298023224,
+      "rewards/chosen": -4.03125,
+      "rewards/margins": 1.484375,
+      "rewards/rejected": -5.53125,
+      "step": 640
+    },
+    {
+      "epoch": 2.07667731629393,
+      "grad_norm": 0.07809236634230764,
+      "learning_rate": 1.710059171597633e-07,
+      "logits/chosen": -1.0625,
+      "logits/rejected": -1.203125,
+      "logps/chosen": -57.75,
+      "logps/rejected": -70.0,
+      "loss": 0.632,
+      "rewards/accuracies": 0.08749999850988388,
+      "rewards/chosen": -3.25,
+      "rewards/margins": 0.9921875,
+      "rewards/rejected": -4.25,
+      "step": 650
+    },
+    {
+      "epoch": 2.108626198083067,
+      "grad_norm": 0.06516317067237166,
+      "learning_rate": 1.650887573964497e-07,
+      "logits/chosen": -0.7265625,
+      "logits/rejected": -0.72265625,
+      "logps/chosen": -76.5,
+      "logps/rejected": -81.0,
+      "loss": 0.6186,
+      "rewards/accuracies": 0.05000000074505806,
+      "rewards/chosen": -5.03125,
+      "rewards/margins": 0.5078125,
+      "rewards/rejected": -5.5625,
+      "step": 660
+    },
+    {
+      "epoch": 2.1405750798722045,
+      "grad_norm": 0.051643150499330034,
+      "learning_rate": 1.591715976331361e-07,
+      "logits/chosen": -0.76171875,
+      "logits/rejected": -0.6796875,
+      "logps/chosen": -79.5,
+      "logps/rejected": -91.5,
+      "loss": 0.6134,
+      "rewards/accuracies": 0.125,
+      "rewards/chosen": -5.28125,
+      "rewards/margins": 1.25,
+      "rewards/rejected": -6.53125,
+      "step": 670
+    },
+    {
+      "epoch": 2.1725239616613417,
+      "grad_norm": 0.08025556916050038,
+      "learning_rate": 1.5325443786982248e-07,
+      "logits/chosen": -0.8671875,
+      "logits/rejected": -0.84765625,
+      "logps/chosen": -79.5,
+      "logps/rejected": -90.5,
+      "loss": 0.6419,
+      "rewards/accuracies": 0.07500000298023224,
+      "rewards/chosen": -5.1875,
+      "rewards/margins": 1.125,
+      "rewards/rejected": -6.28125,
+      "step": 680
+    },
+    {
+      "epoch": 2.2044728434504792,
+      "grad_norm": 0.0576214364801174,
+      "learning_rate": 1.4733727810650885e-07,
+      "logits/chosen": -0.7109375,
+      "logits/rejected": -0.7265625,
+      "logps/chosen": -73.5,
+      "logps/rejected": -88.5,
+      "loss": 0.6011,
+      "rewards/accuracies": 0.125,
+      "rewards/chosen": -4.8125,
+      "rewards/margins": 1.4375,
+      "rewards/rejected": -6.25,
+      "step": 690
+    },
+    {
+      "epoch": 2.236421725239617,
+      "grad_norm": 0.07023261300538382,
+      "learning_rate": 1.4142011834319526e-07,
+      "logits/chosen": -0.703125,
+      "logits/rejected": -0.7265625,
+      "logps/chosen": -80.5,
+      "logps/rejected": -93.5,
+      "loss": 0.6226,
+      "rewards/accuracies": 0.10000000149011612,
+      "rewards/chosen": -5.40625,
+      "rewards/margins": 1.2109375,
+      "rewards/rejected": -6.625,
+      "step": 700
+    },
+    {
+      "epoch": 2.268370607028754,
+      "grad_norm": 2.0705799687047497,
+      "learning_rate": 1.3550295857988164e-07,
+      "logits/chosen": -0.53515625,
+      "logits/rejected": -0.69921875,
+      "logps/chosen": -82.5,
+      "logps/rejected": -97.5,
+      "loss": 0.6126,
+      "rewards/accuracies": 0.11249999701976776,
+      "rewards/chosen": -5.75,
+      "rewards/margins": 1.2421875,
+      "rewards/rejected": -7.0,
+      "step": 710
+    },
+    {
+      "epoch": 2.3003194888178915,
+      "grad_norm": 0.05267573804141954,
+      "learning_rate": 1.2958579881656802e-07,
+      "logits/chosen": -0.5390625,
+      "logits/rejected": -0.416015625,
+      "logps/chosen": -80.0,
+      "logps/rejected": -91.5,
+      "loss": 0.6208,
+      "rewards/accuracies": 0.125,
+      "rewards/chosen": -5.3125,
+      "rewards/margins": 1.2578125,
+      "rewards/rejected": -6.59375,
+      "step": 720
+    },
+    {
+      "epoch": 2.3322683706070286,
+      "grad_norm": 0.06157412465679981,
+      "learning_rate": 1.2366863905325443e-07,
+      "logits/chosen": -0.333984375,
+      "logits/rejected": -0.330078125,
+      "logps/chosen": -93.0,
+      "logps/rejected": -100.0,
+      "loss": 0.6077,
+      "rewards/accuracies": 0.07500000298023224,
+      "rewards/chosen": -6.6875,
+      "rewards/margins": 0.69921875,
+      "rewards/rejected": -7.375,
+      "step": 730
+    },
+    {
+      "epoch": 2.364217252396166,
+      "grad_norm": 0.5344640945556689,
+      "learning_rate": 1.1775147928994082e-07,
+      "logits/chosen": -0.36328125,
+      "logits/rejected": -0.359375,
+      "logps/chosen": -84.0,
+      "logps/rejected": -92.0,
+      "loss": 0.6148,
+      "rewards/accuracies": 0.0625,
+      "rewards/chosen": -5.875,
+      "rewards/margins": 0.75390625,
+      "rewards/rejected": -6.625,
+      "step": 740
+    },
+    {
+      "epoch": 2.3961661341853033,
+      "grad_norm": 0.062456246205116854,
+      "learning_rate": 1.1183431952662721e-07,
+      "logits/chosen": -0.5625,
+      "logits/rejected": -0.388671875,
+      "logps/chosen": -87.0,
+      "logps/rejected": -99.0,
+      "loss": 0.6118,
+      "rewards/accuracies": 0.13750000298023224,
+      "rewards/chosen": -5.90625,
+      "rewards/margins": 1.40625,
+      "rewards/rejected": -7.3125,
+      "step": 750
+    },
+    {
+      "epoch": 2.428115015974441,
+      "grad_norm": 0.04969846393608146,
+      "learning_rate": 1.059171597633136e-07,
+      "logits/chosen": -0.40625,
+      "logits/rejected": -0.3515625,
+      "logps/chosen": -86.0,
+      "logps/rejected": -106.5,
+      "loss": 0.6032,
+      "rewards/accuracies": 0.17499999701976776,
+      "rewards/chosen": -5.96875,
+      "rewards/margins": 2.015625,
+      "rewards/rejected": -8.0,
+      "step": 760
+    },
+    {
+      "epoch": 2.460063897763578,
+      "grad_norm": 0.07639092395372227,
+      "learning_rate": 1e-07,
+      "logits/chosen": -0.357421875,
+      "logits/rejected": -0.5,
+      "logps/chosen": -84.5,
+      "logps/rejected": -107.5,
+      "loss": 0.6055,
+      "rewards/accuracies": 0.15000000596046448,
+      "rewards/chosen": -5.875,
+      "rewards/margins": 2.1875,
+      "rewards/rejected": -8.0625,
+      "step": 770
+    },
+    {
+      "epoch": 2.4920127795527156,
+      "grad_norm": 0.07447444348205871,
+      "learning_rate": 9.408284023668639e-08,
+      "logits/chosen": -0.5078125,
+      "logits/rejected": -0.61328125,
+      "logps/chosen": -81.0,
+      "logps/rejected": -103.5,
+      "loss": 0.6054,
+      "rewards/accuracies": 0.16249999403953552,
+      "rewards/chosen": -5.46875,
+      "rewards/margins": 2.09375,
+      "rewards/rejected": -7.5625,
+      "step": 780
+    },
+    {
+      "epoch": 2.523961661341853,
+      "grad_norm": 0.10040200439902261,
+      "learning_rate": 8.816568047337278e-08,
+      "logits/chosen": -0.578125,
+      "logits/rejected": -0.482421875,
+      "logps/chosen": -74.0,
+      "logps/rejected": -89.0,
+      "loss": 0.6253,
+      "rewards/accuracies": 0.125,
+      "rewards/chosen": -4.8125,
+      "rewards/margins": 1.453125,
+      "rewards/rejected": -6.28125,
+      "step": 790
+    },
+    {
+      "epoch": 2.5559105431309903,
+      "grad_norm": 0.23663153627221573,
+      "learning_rate": 8.224852071005916e-08,
+      "logits/chosen": -0.482421875,
+      "logits/rejected": -0.5,
+      "logps/chosen": -76.5,
+      "logps/rejected": -93.5,
+      "loss": 0.5994,
+      "rewards/accuracies": 0.125,
+      "rewards/chosen": -5.09375,
+      "rewards/margins": 1.6796875,
+      "rewards/rejected": -6.78125,
+      "step": 800
+    },
+    {
+      "epoch": 2.587859424920128,
+      "grad_norm": 1.1904000627005848,
+      "learning_rate": 7.633136094674555e-08,
+      "logits/chosen": -0.61328125,
+      "logits/rejected": -0.55078125,
+      "logps/chosen": -81.0,
+      "logps/rejected": -97.0,
+      "loss": 0.6114,
+      "rewards/accuracies": 0.15000000596046448,
+      "rewards/chosen": -5.5,
+      "rewards/margins": 1.5546875,
+      "rewards/rejected": -7.0625,
+      "step": 810
+    },
+    {
+      "epoch": 2.619808306709265,
+      "grad_norm": 0.06181041676045537,
+      "learning_rate": 7.041420118343195e-08,
+      "logits/chosen": -0.5625,
+      "logits/rejected": -0.58203125,
+      "logps/chosen": -85.5,
+      "logps/rejected": -99.5,
+      "loss": 0.6229,
+      "rewards/accuracies": 0.125,
+      "rewards/chosen": -5.78125,
+      "rewards/margins": 1.421875,
+      "rewards/rejected": -7.21875,
+      "step": 820
+    },
+    {
+      "epoch": 2.6517571884984026,
+      "grad_norm": 0.06412244004564162,
+      "learning_rate": 6.449704142011835e-08,
+      "logits/chosen": -0.68359375,
+      "logits/rejected": -0.57421875,
+      "logps/chosen": -73.0,
+      "logps/rejected": -86.0,
+      "loss": 0.6204,
+      "rewards/accuracies": 0.15000000596046448,
+      "rewards/chosen": -4.59375,
+      "rewards/margins": 1.375,
+      "rewards/rejected": -5.96875,
+      "step": 830
+    },
+    {
+      "epoch": 2.68370607028754,
+      "grad_norm": 18.114477819614837,
+      "learning_rate": 5.857988165680473e-08,
+      "logits/chosen": -0.71875,
+      "logits/rejected": -0.54296875,
+      "logps/chosen": -87.5,
+      "logps/rejected": -125.5,
+      "loss": 0.6208,
+      "rewards/accuracies": 0.11249999701976776,
+      "rewards/chosen": -5.875,
+      "rewards/margins": 3.96875,
+      "rewards/rejected": -9.8125,
+      "step": 840
+    },
+    {
+      "epoch": 2.7156549520766773,
+      "grad_norm": 0.0580429780200346,
+      "learning_rate": 5.266272189349112e-08,
+      "logits/chosen": -0.5078125,
+      "logits/rejected": -0.494140625,
+      "logps/chosen": -84.0,
+      "logps/rejected": -91.0,
+      "loss": 0.6254,
+      "rewards/accuracies": 0.07500000298023224,
+      "rewards/chosen": -5.8125,
+      "rewards/margins": 0.6640625,
+      "rewards/rejected": -6.5,
+      "step": 850
+    },
+    {
+      "epoch": 2.747603833865815,
+      "grad_norm": 0.07862204388717929,
+      "learning_rate": 4.674556213017751e-08,
+      "logits/chosen": -0.52734375,
+      "logits/rejected": -0.5390625,
+      "logps/chosen": -82.0,
+      "logps/rejected": -100.0,
+      "loss": 0.5989,
+      "rewards/accuracies": 0.13750000298023224,
+      "rewards/chosen": -5.59375,
+      "rewards/margins": 1.6875,
+      "rewards/rejected": -7.28125,
+      "step": 860
+    },
+    {
+      "epoch": 2.779552715654952,
+      "grad_norm": 1.3694461818497943,
+      "learning_rate": 4.082840236686391e-08,
+      "logits/chosen": -0.546875,
+      "logits/rejected": -0.5234375,
+      "logps/chosen": -84.5,
+      "logps/rejected": -93.5,
+      "loss": 0.6144,
+      "rewards/accuracies": 0.07500000298023224,
+      "rewards/chosen": -5.84375,
+      "rewards/margins": 0.87890625,
+      "rewards/rejected": -6.71875,
+      "step": 870
+    },
+    {
+      "epoch": 2.8115015974440896,
+      "grad_norm": 0.09117311902195645,
+      "learning_rate": 3.4911242603550294e-08,
+      "logits/chosen": -0.478515625,
+      "logits/rejected": -0.44140625,
+      "logps/chosen": -85.0,
+      "logps/rejected": -95.5,
+      "loss": 0.6142,
+      "rewards/accuracies": 0.10000000149011612,
+      "rewards/chosen": -5.9375,
+      "rewards/margins": 1.046875,
+      "rewards/rejected": -6.96875,
+      "step": 880
+    },
+    {
+      "epoch": 2.8434504792332267,
+      "grad_norm": 6.5275431311324645,
+      "learning_rate": 2.8994082840236687e-08,
+      "logits/chosen": -0.5234375,
+      "logits/rejected": -0.416015625,
+      "logps/chosen": -92.0,
+      "logps/rejected": -106.0,
+      "loss": 0.617,
+      "rewards/accuracies": 0.125,
+      "rewards/chosen": -6.4375,
+      "rewards/margins": 1.484375,
+      "rewards/rejected": -7.9375,
+      "step": 890
+    },
+    {
+      "epoch": 2.8753993610223643,
+      "grad_norm": 0.06494665152999182,
+      "learning_rate": 2.3076923076923076e-08,
+      "logits/chosen": -0.349609375,
+      "logits/rejected": -0.37890625,
+      "logps/chosen": -84.0,
+      "logps/rejected": -97.0,
+      "loss": 0.5968,
+      "rewards/accuracies": 0.125,
+      "rewards/chosen": -5.90625,
+      "rewards/margins": 1.2421875,
+      "rewards/rejected": -7.15625,
+      "step": 900
+    },
+    {
+      "epoch": 2.9073482428115014,
+      "grad_norm": 0.10041685712641052,
+      "learning_rate": 1.7159763313609465e-08,
+      "logits/chosen": -0.392578125,
+      "logits/rejected": -0.32421875,
+      "logps/chosen": -82.5,
+      "logps/rejected": -97.0,
+      "loss": 0.601,
+      "rewards/accuracies": 0.125,
+      "rewards/chosen": -5.6875,
+      "rewards/margins": 1.4140625,
+      "rewards/rejected": -7.09375,
+      "step": 910
+    },
+    {
+      "epoch": 2.939297124600639,
+      "grad_norm": 0.07270087893217783,
+      "learning_rate": 1.1242603550295858e-08,
+      "logits/chosen": -0.53515625,
+      "logits/rejected": -0.474609375,
+      "logps/chosen": -81.5,
+      "logps/rejected": -101.5,
+      "loss": 0.6098,
+      "rewards/accuracies": 0.17499999701976776,
+      "rewards/chosen": -5.4375,
+      "rewards/margins": 2.015625,
+      "rewards/rejected": -7.46875,
+      "step": 920
+    },
+    {
+      "epoch": 2.9712460063897765,
+      "grad_norm": 0.05599532699973568,
+      "learning_rate": 5.325443786982248e-09,
+      "logits/chosen": -0.376953125,
+      "logits/rejected": -0.314453125,
+      "logps/chosen": -88.5,
+      "logps/rejected": -98.0,
+      "loss": 0.6397,
+      "rewards/accuracies": 0.08749999850988388,
+      "rewards/chosen": -6.25,
+      "rewards/margins": 0.94921875,
+      "rewards/rejected": -7.1875,
+      "step": 930
+    },
+    {
+      "epoch": 3.0,
+      "eval_logits/chosen": -0.375,
+      "eval_logits/rejected": -0.40625,
+      "eval_logps/chosen": -85.0,
+      "eval_logps/rejected": -85.5,
+      "eval_loss": 0.7239843606948853,
+      "eval_rewards/accuracies": 0.0357142873108387,
+      "eval_rewards/chosen": -5.96875,
+      "eval_rewards/margins": 0.046142578125,
+      "eval_rewards/rejected": -6.0,
+      "eval_runtime": 15.3339,
+      "eval_samples_per_second": 13.043,
+      "eval_steps_per_second": 0.457,
+      "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": {}
     }