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:a348a22c4e7292a63c81cfc0fcc1b8ac3dae85423a483a0702496957d13b6b95
+size 13476836524

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 sha256:1d1fa4db8c10c30bd860a7eaef1f71d3c5747deb313934742d2744935d062557
+size 13476836524

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 sha256:4b51bda3e9193367ef28889dd6408835151474efb556046c8f933c6ca91a5431
+size 13476836524

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 sha256:6beeba792c0b122547d06c33b304f468ea691c48a6de2d30ffd2663bad9f14c4
+size 13476836524

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 sha256:0f04249e1b8fe1c2cdc217cf32f7eea80a19cb28668041a7568bda12381df332
+size 150693

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 sha256:2a8e7204e992a56a415d70b4b33a5427de747e33a8410a8d2e57b2d3760f41a3
+size 150693

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:b82bc0de7fa407ebf2e63d8203a63e23579c6341814abb2367c0d8f0fb95b0c2
+size 150693

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:f568cb4b0ff22d0f18efb7a633cdedc227f3a80be6b0c1fcc6c09a4985216c2c
+size 150693

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:f5904e952701479e892c2d17a4d0b611f4ed302daf4b6a1aba2eaf266f860079
+oid sha256:3bba1df728f0a13c32473530dfdf8f2b9e407ef34cc781c3f369d508860f0ac6
 size 4938985352

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

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8a2daac32f9ba6f35b798552797ed4b27987e5b51730c584e2939993e47216dd
+oid sha256:11b000d5ce7e244db5581aa54cbceba6b8330bae9c6128c79cb7cc8b33fe3425
 size 4947390880

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

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3c2d3fdd1381913365ea0afbde72a0f74775bce063d67ec06dcf7ecb72d472be
+oid sha256:8848153f15166b9cad0472ba12cd8877ab2aef7f2e79e9af3a0a62a3742c4156
 size 3590488816

last-checkpoint/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f1ce770862b76ceecb3453a5d0b9ef65311eab0f6a60f52997726a89d718de08
+oid sha256:2307c03867cef25b5028feb9a23f80e784b9af9a615de13ddca560a6a90fb593
 size 15024

last-checkpoint/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2bcae6573442578b752be3e988d4ccb38056b45c31b5c02eb579ec6cebcfa62e
+oid sha256:50218cfaecdd818354e567b7167c13899e3b42297e7d8f58bd7e732cfa547800
 size 15024

last-checkpoint/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:398ff45072fee2975e88b9e078f915103d75d94bc08753303cf855b915973623
+oid sha256:9814a66b49861f5495b06dae3be12ddf7185b88e2cae1fb808ca9efd99d5807f
 size 15024

last-checkpoint/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7f4b210bc832d4c648e3bee8c21dca26e5a8b365d6ec90c638062005a052e57b
+oid sha256:a7be93040a65e0a29975f6c70b94418e1fdf88423a50c58aa572141d3c92fbfc
 size 15024

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6b9aad3a748e15719e20120f0a760f22686338e176ee3f1c011b6dab2f3ab19b
+oid sha256:f7725b44c69725321786f1e58dccd7ea4d3ae5794ea47bd49c0d4a139aec266d
 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": 16.39,
       "eval_steps_per_second": 0.574,
       "step": 313
+    },
+    {
+      "epoch": 1.0223642172523961,
+      "grad_norm": 15.216502023078874,
+      "learning_rate": 3.662721893491124e-07,
+      "logits/chosen": -1.578125,
+      "logits/rejected": -1.59375,
+      "logps/chosen": -230.0,
+      "logps/rejected": -262.0,
+      "loss": 0.4383,
+      "rewards/accuracies": 0.8500000238418579,
+      "rewards/chosen": 0.66015625,
+      "rewards/margins": 1.0546875,
+      "rewards/rejected": -0.392578125,
+      "step": 320
+    },
+    {
+      "epoch": 1.0543130990415335,
+      "grad_norm": 16.390892234984523,
+      "learning_rate": 3.603550295857988e-07,
+      "logits/chosen": -1.4921875,
+      "logits/rejected": -1.59375,
+      "logps/chosen": -232.0,
+      "logps/rejected": -276.0,
+      "loss": 0.3885,
+      "rewards/accuracies": 0.9375,
+      "rewards/chosen": 0.6328125,
+      "rewards/margins": 1.2578125,
+      "rewards/rejected": -0.625,
+      "step": 330
+    },
+    {
+      "epoch": 1.0862619808306708,
+      "grad_norm": 14.29774575781621,
+      "learning_rate": 3.544378698224852e-07,
+      "logits/chosen": -1.5390625,
+      "logits/rejected": -1.5546875,
+      "logps/chosen": -228.0,
+      "logps/rejected": -268.0,
+      "loss": 0.3978,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": 0.63671875,
+      "rewards/margins": 1.2265625,
+      "rewards/rejected": -0.58984375,
+      "step": 340
+    },
+    {
+      "epoch": 1.1182108626198084,
+      "grad_norm": 16.80859709996563,
+      "learning_rate": 3.485207100591716e-07,
+      "logits/chosen": -1.5390625,
+      "logits/rejected": -1.5078125,
+      "logps/chosen": -236.0,
+      "logps/rejected": -268.0,
+      "loss": 0.3711,
+      "rewards/accuracies": 0.8374999761581421,
+      "rewards/chosen": 0.50390625,
+      "rewards/margins": 1.0859375,
+      "rewards/rejected": -0.58203125,
+      "step": 350
+    },
+    {
+      "epoch": 1.1501597444089458,
+      "grad_norm": 17.099448590902114,
+      "learning_rate": 3.42603550295858e-07,
+      "logits/chosen": -1.53125,
+      "logits/rejected": -1.5625,
+      "logps/chosen": -228.0,
+      "logps/rejected": -262.0,
+      "loss": 0.3884,
+      "rewards/accuracies": 0.800000011920929,
+      "rewards/chosen": 0.515625,
+      "rewards/margins": 1.0546875,
+      "rewards/rejected": -0.54296875,
+      "step": 360
+    },
+    {
+      "epoch": 1.182108626198083,
+      "grad_norm": 15.964788390337974,
+      "learning_rate": 3.366863905325444e-07,
+      "logits/chosen": -1.5625,
+      "logits/rejected": -1.5625,
+      "logps/chosen": -234.0,
+      "logps/rejected": -266.0,
+      "loss": 0.3964,
+      "rewards/accuracies": 0.824999988079071,
+      "rewards/chosen": 0.5546875,
+      "rewards/margins": 0.97265625,
+      "rewards/rejected": -0.416015625,
+      "step": 370
+    },
+    {
+      "epoch": 1.2140575079872205,
+      "grad_norm": 17.10451165088142,
+      "learning_rate": 3.3076923076923075e-07,
+      "logits/chosen": -1.5546875,
+      "logits/rejected": -1.5703125,
+      "logps/chosen": -232.0,
+      "logps/rejected": -260.0,
+      "loss": 0.3918,
+      "rewards/accuracies": 0.762499988079071,
+      "rewards/chosen": 0.51953125,
+      "rewards/margins": 1.015625,
+      "rewards/rejected": -0.494140625,
+      "step": 380
+    },
+    {
+      "epoch": 1.2460063897763578,
+      "grad_norm": 16.51862951602937,
+      "learning_rate": 3.2485207100591716e-07,
+      "logits/chosen": -1.546875,
+      "logits/rejected": -1.6328125,
+      "logps/chosen": -228.0,
+      "logps/rejected": -276.0,
+      "loss": 0.3695,
+      "rewards/accuracies": 0.8999999761581421,
+      "rewards/chosen": 0.8125,
+      "rewards/margins": 1.6953125,
+      "rewards/rejected": -0.88671875,
+      "step": 390
+    },
+    {
+      "epoch": 1.2779552715654952,
+      "grad_norm": 19.593966068797595,
+      "learning_rate": 3.189349112426035e-07,
+      "logits/chosen": -1.5859375,
+      "logits/rejected": -1.625,
+      "logps/chosen": -235.0,
+      "logps/rejected": -276.0,
+      "loss": 0.4045,
+      "rewards/accuracies": 0.800000011920929,
+      "rewards/chosen": 0.5234375,
+      "rewards/margins": 1.125,
+      "rewards/rejected": -0.59765625,
+      "step": 400
+    },
+    {
+      "epoch": 1.3099041533546325,
+      "grad_norm": 19.821206837758886,
+      "learning_rate": 3.130177514792899e-07,
+      "logits/chosen": -1.546875,
+      "logits/rejected": -1.6015625,
+      "logps/chosen": -239.0,
+      "logps/rejected": -266.0,
+      "loss": 0.4309,
+      "rewards/accuracies": 0.8500000238418579,
+      "rewards/chosen": 0.427734375,
+      "rewards/margins": 1.1640625,
+      "rewards/rejected": -0.734375,
+      "step": 410
+    },
+    {
+      "epoch": 1.34185303514377,
+      "grad_norm": 17.457629857186152,
+      "learning_rate": 3.071005917159763e-07,
+      "logits/chosen": -1.625,
+      "logits/rejected": -1.609375,
+      "logps/chosen": -243.0,
+      "logps/rejected": -270.0,
+      "loss": 0.3913,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": 0.4609375,
+      "rewards/margins": 1.1796875,
+      "rewards/rejected": -0.72265625,
+      "step": 420
+    },
+    {
+      "epoch": 1.3738019169329074,
+      "grad_norm": 16.963757931748397,
+      "learning_rate": 3.011834319526627e-07,
+      "logits/chosen": -1.578125,
+      "logits/rejected": -1.609375,
+      "logps/chosen": -235.0,
+      "logps/rejected": -270.0,
+      "loss": 0.3808,
+      "rewards/accuracies": 0.8374999761581421,
+      "rewards/chosen": 0.41796875,
+      "rewards/margins": 1.1953125,
+      "rewards/rejected": -0.78125,
+      "step": 430
+    },
+    {
+      "epoch": 1.4057507987220448,
+      "grad_norm": 16.933512859408392,
+      "learning_rate": 2.952662721893491e-07,
+      "logits/chosen": -1.546875,
+      "logits/rejected": -1.578125,
+      "logps/chosen": -238.0,
+      "logps/rejected": -262.0,
+      "loss": 0.4261,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": 0.40234375,
+      "rewards/margins": 1.0,
+      "rewards/rejected": -0.6015625,
+      "step": 440
+    },
+    {
+      "epoch": 1.4376996805111821,
+      "grad_norm": 17.582118451476234,
+      "learning_rate": 2.893491124260355e-07,
+      "logits/chosen": -1.625,
+      "logits/rejected": -1.6015625,
+      "logps/chosen": -235.0,
+      "logps/rejected": -268.0,
+      "loss": 0.3921,
+      "rewards/accuracies": 0.800000011920929,
+      "rewards/chosen": 0.435546875,
+      "rewards/margins": 1.0546875,
+      "rewards/rejected": -0.62109375,
+      "step": 450
+    },
+    {
+      "epoch": 1.4696485623003195,
+      "grad_norm": 14.599835012516651,
+      "learning_rate": 2.834319526627219e-07,
+      "logits/chosen": -1.5546875,
+      "logits/rejected": -1.5859375,
+      "logps/chosen": -234.0,
+      "logps/rejected": -266.0,
+      "loss": 0.357,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": 0.353515625,
+      "rewards/margins": 1.1796875,
+      "rewards/rejected": -0.828125,
+      "step": 460
+    },
+    {
+      "epoch": 1.5015974440894568,
+      "grad_norm": 17.01992762798607,
+      "learning_rate": 2.7751479289940824e-07,
+      "logits/chosen": -1.5625,
+      "logits/rejected": -1.625,
+      "logps/chosen": -238.0,
+      "logps/rejected": -272.0,
+      "loss": 0.3465,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": 0.232421875,
+      "rewards/margins": 1.1640625,
+      "rewards/rejected": -0.93359375,
+      "step": 470
+    },
+    {
+      "epoch": 1.5335463258785942,
+      "grad_norm": 16.926918987794753,
+      "learning_rate": 2.715976331360947e-07,
+      "logits/chosen": -1.625,
+      "logits/rejected": -1.578125,
+      "logps/chosen": -245.0,
+      "logps/rejected": -266.0,
+      "loss": 0.3938,
+      "rewards/accuracies": 0.762499988079071,
+      "rewards/chosen": 0.2197265625,
+      "rewards/margins": 1.0078125,
+      "rewards/rejected": -0.78515625,
+      "step": 480
+    },
+    {
+      "epoch": 1.5654952076677318,
+      "grad_norm": 19.019339456655477,
+      "learning_rate": 2.6568047337278106e-07,
+      "logits/chosen": -1.6328125,
+      "logits/rejected": -1.625,
+      "logps/chosen": -232.0,
+      "logps/rejected": -272.0,
+      "loss": 0.3621,
+      "rewards/accuracies": 0.887499988079071,
+      "rewards/chosen": 0.369140625,
+      "rewards/margins": 1.3125,
+      "rewards/rejected": -0.9453125,
+      "step": 490
+    },
+    {
+      "epoch": 1.5974440894568689,
+      "grad_norm": 19.071902381545915,
+      "learning_rate": 2.5976331360946746e-07,
+      "logits/chosen": -1.5859375,
+      "logits/rejected": -1.6015625,
+      "logps/chosen": -231.0,
+      "logps/rejected": -266.0,
+      "loss": 0.4204,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": 0.314453125,
+      "rewards/margins": 1.1796875,
+      "rewards/rejected": -0.8671875,
+      "step": 500
+    },
+    {
+      "epoch": 1.6293929712460065,
+      "grad_norm": 15.876124711628384,
+      "learning_rate": 2.538461538461538e-07,
+      "logits/chosen": -1.6171875,
+      "logits/rejected": -1.609375,
+      "logps/chosen": -236.0,
+      "logps/rejected": -262.0,
+      "loss": 0.3832,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": 0.365234375,
+      "rewards/margins": 1.2109375,
+      "rewards/rejected": -0.84375,
+      "step": 510
+    },
+    {
+      "epoch": 1.6613418530351438,
+      "grad_norm": 14.900626138023746,
+      "learning_rate": 2.479289940828402e-07,
+      "logits/chosen": -1.578125,
+      "logits/rejected": -1.5625,
+      "logps/chosen": -232.0,
+      "logps/rejected": -268.0,
+      "loss": 0.3459,
+      "rewards/accuracies": 0.8500000238418579,
+      "rewards/chosen": 0.458984375,
+      "rewards/margins": 1.46875,
+      "rewards/rejected": -1.015625,
+      "step": 520
+    },
+    {
+      "epoch": 1.6932907348242812,
+      "grad_norm": 19.494063656117618,
+      "learning_rate": 2.420118343195266e-07,
+      "logits/chosen": -1.6015625,
+      "logits/rejected": -1.625,
+      "logps/chosen": -230.0,
+      "logps/rejected": -278.0,
+      "loss": 0.3752,
+      "rewards/accuracies": 0.7875000238418579,
+      "rewards/chosen": 0.3671875,
+      "rewards/margins": 1.328125,
+      "rewards/rejected": -0.96484375,
+      "step": 530
+    },
+    {
+      "epoch": 1.7252396166134185,
+      "grad_norm": 15.766044347009432,
+      "learning_rate": 2.36094674556213e-07,
+      "logits/chosen": -1.59375,
+      "logits/rejected": -1.6171875,
+      "logps/chosen": -231.0,
+      "logps/rejected": -270.0,
+      "loss": 0.3902,
+      "rewards/accuracies": 0.8500000238418579,
+      "rewards/chosen": 0.392578125,
+      "rewards/margins": 1.3359375,
+      "rewards/rejected": -0.9375,
+      "step": 540
+    },
+    {
+      "epoch": 1.7571884984025559,
+      "grad_norm": 17.890212932549957,
+      "learning_rate": 2.301775147928994e-07,
+      "logits/chosen": -1.59375,
+      "logits/rejected": -1.6328125,
+      "logps/chosen": -241.0,
+      "logps/rejected": -256.0,
+      "loss": 0.3833,
+      "rewards/accuracies": 0.800000011920929,
+      "rewards/chosen": 0.384765625,
+      "rewards/margins": 0.9921875,
+      "rewards/rejected": -0.609375,
+      "step": 550
+    },
+    {
+      "epoch": 1.7891373801916934,
+      "grad_norm": 17.977480288742196,
+      "learning_rate": 2.242603550295858e-07,
+      "logits/chosen": -1.578125,
+      "logits/rejected": -1.625,
+      "logps/chosen": -239.0,
+      "logps/rejected": -270.0,
+      "loss": 0.3717,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": 0.259765625,
+      "rewards/margins": 1.078125,
+      "rewards/rejected": -0.8203125,
+      "step": 560
+    },
+    {
+      "epoch": 1.8210862619808306,
+      "grad_norm": 17.31063553493517,
+      "learning_rate": 2.1834319526627217e-07,
+      "logits/chosen": -1.6015625,
+      "logits/rejected": -1.59375,
+      "logps/chosen": -233.0,
+      "logps/rejected": -280.0,
+      "loss": 0.3695,
+      "rewards/accuracies": 0.9125000238418579,
+      "rewards/chosen": 0.283203125,
+      "rewards/margins": 1.28125,
+      "rewards/rejected": -0.99609375,
+      "step": 570
+    },
+    {
+      "epoch": 1.8530351437699681,
+      "grad_norm": 21.27064447874725,
+      "learning_rate": 2.1242603550295858e-07,
+      "logits/chosen": -1.578125,
+      "logits/rejected": -1.6015625,
+      "logps/chosen": -242.0,
+      "logps/rejected": -276.0,
+      "loss": 0.3861,
+      "rewards/accuracies": 0.7875000238418579,
+      "rewards/chosen": 0.234375,
+      "rewards/margins": 1.0546875,
+      "rewards/rejected": -0.8203125,
+      "step": 580
+    },
+    {
+      "epoch": 1.8849840255591053,
+      "grad_norm": 19.92306872555079,
+      "learning_rate": 2.0650887573964496e-07,
+      "logits/chosen": -1.5859375,
+      "logits/rejected": -1.5859375,
+      "logps/chosen": -235.0,
+      "logps/rejected": -264.0,
+      "loss": 0.3873,
+      "rewards/accuracies": 0.7875000238418579,
+      "rewards/chosen": 0.37890625,
+      "rewards/margins": 1.1015625,
+      "rewards/rejected": -0.72265625,
+      "step": 590
+    },
+    {
+      "epoch": 1.9169329073482428,
+      "grad_norm": 22.677693548477258,
+      "learning_rate": 2.0059171597633133e-07,
+      "logits/chosen": -1.6171875,
+      "logits/rejected": -1.6171875,
+      "logps/chosen": -240.0,
+      "logps/rejected": -258.0,
+      "loss": 0.4016,
+      "rewards/accuracies": 0.800000011920929,
+      "rewards/chosen": 0.2314453125,
+      "rewards/margins": 0.984375,
+      "rewards/rejected": -0.75390625,
+      "step": 600
+    },
+    {
+      "epoch": 1.9488817891373802,
+      "grad_norm": 21.51355955572424,
+      "learning_rate": 1.9467455621301774e-07,
+      "logits/chosen": -1.578125,
+      "logits/rejected": -1.671875,
+      "logps/chosen": -234.0,
+      "logps/rejected": -266.0,
+      "loss": 0.375,
+      "rewards/accuracies": 0.800000011920929,
+      "rewards/chosen": 0.25390625,
+      "rewards/margins": 1.0859375,
+      "rewards/rejected": -0.83203125,
+      "step": 610
+    },
+    {
+      "epoch": 1.9808306709265175,
+      "grad_norm": 14.74606499354998,
+      "learning_rate": 1.8875739644970412e-07,
+      "logits/chosen": -1.578125,
+      "logits/rejected": -1.59375,
+      "logps/chosen": -243.0,
+      "logps/rejected": -268.0,
+      "loss": 0.3809,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": 0.03759765625,
+      "rewards/margins": 1.1328125,
+      "rewards/rejected": -1.09375,
+      "step": 620
+    },
+    {
+      "epoch": 2.0,
+      "eval_logits/chosen": -1.6171875,
+      "eval_logits/rejected": -1.59375,
+      "eval_logps/chosen": -244.0,
+      "eval_logps/rejected": -260.0,
+      "eval_loss": 0.5935156345367432,
+      "eval_rewards/accuracies": 0.6428571343421936,
+      "eval_rewards/chosen": -0.0556640625,
+      "eval_rewards/margins": 0.63671875,
+      "eval_rewards/rejected": -0.69140625,
+      "eval_runtime": 12.2896,
+      "eval_samples_per_second": 16.274,
+      "eval_steps_per_second": 0.57,
+      "step": 626
     }
   ],
   "logging_steps": 10,