SFT checkpoint step 58

checkpoints/checkpoint-58/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:634907b87fad29c41e1d4d85097416b68624027cd97eb45bc31006a260832795
+oid sha256:fc857bcd5904c962a3383bf98b0b0e2dbdb8205d4292258e7ab24adedcd170b7
 size 204500912

checkpoints/checkpoint-58/mtp_head.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4596cd452e7430fcbb46b6be680fc27b050513f20b4211b381fbdc6c4a951488
-size 40912203
+oid sha256:f293bb634d6dfe02eb83f0be38ee50a559118344add09a5875a34c252af49c7f
+size 81817931

checkpoints/checkpoint-58/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4eb50e46b8936a621fde73105836e718f934d55c1a76fee76f3c39e6451052e3
-size 104062795
+oid sha256:4e13095a45e7433c53c172add8a9d597f38278b60b51dc4a6fc1b3d90174ed09
+size 620569547

checkpoints/checkpoint-58/scaler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8adfb47887974e29804b6fe7fb969b1afa8105838a60d173990ae3c0be86b6f0
+oid sha256:72b216d080578100bd57ee7cdb48421e348ca42dd3c8025e9877a55f62e8737f
 size 1383

checkpoints/checkpoint-58/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a5e9a21fee0ed695b918359d73521fd31afdc5745d9a5621dafc5600643d2c22
+oid sha256:645097abd69992527144f761706f73c4c082ca049e8ce9eff30ed02004535f77
 size 1465

checkpoints/checkpoint-58/time_state.json

@@ -1 +1 @@
-{"consumed_seconds": 381.40962314605713, "global_step": 58}
+{"consumed_seconds": 510.2071752548218, "global_step": 58}

checkpoints/checkpoint-58/trainer_state.json

@@ -13,215 +13,215 @@
       "epoch": 0.017391304347826087,
       "grad_norm": NaN,
       "learning_rate": 0.0,
-      "loss": 2.3871500492095947,
+      "loss": 2.5696372985839844,
       "num_input_tokens_seen": 1494,
       "num_tokens": 1494.0,
       "step": 1,
-      "train_runtime": 18.9586,
-      "train_tokens_per_second": 78.803
+      "train_runtime": 36.3516,
+      "train_tokens_per_second": 41.099
     },
     {
       "epoch": 0.06956521739130435,
-      "eval_loss": 5.106939315795898,
+      "eval_loss": 5.31978178024292,
       "eval_num_tokens": 4980.0,
-      "eval_runtime": 13.7855,
-      "eval_samples_per_second": 2.176,
-      "eval_steps_per_second": 0.58,
+      "eval_runtime": 28.3903,
+      "eval_samples_per_second": 1.057,
+      "eval_steps_per_second": 0.282,
       "num_input_tokens_seen": 4980,
       "step": 4
     },
     {
       "epoch": 0.1391304347826087,
-      "eval_loss": 5.106939315795898,
+      "eval_loss": 5.31978178024292,
       "eval_num_tokens": 9004.0,
-      "eval_runtime": 2.0516,
-      "eval_samples_per_second": 14.623,
-      "eval_steps_per_second": 3.899,
+      "eval_runtime": 3.1825,
+      "eval_samples_per_second": 9.427,
+      "eval_steps_per_second": 2.514,
       "num_input_tokens_seen": 9004,
       "step": 8
     },
     {
       "epoch": 0.17391304347826086,
-      "grad_norm": 22.719650268554688,
-      "learning_rate": 7.5e-05,
-      "loss": 2.455470191107856,
+      "grad_norm": 28.139263153076172,
+      "learning_rate": 0.0,
+      "loss": 2.5477417839898004,
       "num_input_tokens_seen": 11145,
       "num_tokens": 11145.0,
       "step": 10,
-      "train_runtime": 96.6319,
-      "train_tokens_per_second": 115.335
+      "train_runtime": 137.8118,
+      "train_tokens_per_second": 80.871
     },
     {
       "epoch": 0.20869565217391303,
-      "eval_loss": 1.0586106777191162,
+      "eval_loss": 1.8169835805892944,
       "eval_num_tokens": 13293.0,
-      "eval_runtime": 2.0908,
-      "eval_samples_per_second": 14.349,
-      "eval_steps_per_second": 3.826,
+      "eval_runtime": 3.1498,
+      "eval_samples_per_second": 9.525,
+      "eval_steps_per_second": 2.54,
       "num_input_tokens_seen": 13293,
       "step": 12
     },
     {
       "epoch": 0.2782608695652174,
-      "eval_loss": 0.4196360409259796,
+      "eval_loss": 0.6515728831291199,
       "eval_num_tokens": 17742.0,
-      "eval_runtime": 2.1417,
-      "eval_samples_per_second": 14.008,
-      "eval_steps_per_second": 3.735,
+      "eval_runtime": 3.1545,
+      "eval_samples_per_second": 9.51,
+      "eval_steps_per_second": 2.536,
       "num_input_tokens_seen": 17742,
       "step": 16
     },
     {
       "epoch": 0.34782608695652173,
-      "grad_norm": 3.528010606765747,
+      "grad_norm": 20.463083267211914,
       "learning_rate": 0.0001425726650926814,
-      "loss": 0.5043737411499023,
+      "loss": 0.7741940021514893,
       "num_input_tokens_seen": 20264,
       "num_tokens": 20264.0,
       "step": 20,
-      "train_runtime": 166.3694,
-      "train_tokens_per_second": 121.801
+      "train_runtime": 223.0367,
+      "train_tokens_per_second": 90.855
     },
     {
       "epoch": 0.34782608695652173,
-      "eval_loss": 0.33024391531944275,
+      "eval_loss": 0.3478561341762543,
       "eval_num_tokens": 20264.0,
-      "eval_runtime": 2.0878,
-      "eval_samples_per_second": 14.369,
-      "eval_steps_per_second": 3.832,
+      "eval_runtime": 3.1698,
+      "eval_samples_per_second": 9.464,
+      "eval_steps_per_second": 2.524,
       "num_input_tokens_seen": 20264,
       "step": 20
     },
     {
       "epoch": 0.41739130434782606,
-      "eval_loss": 0.26133954524993896,
+      "eval_loss": 0.3752416670322418,
       "eval_num_tokens": 24990.0,
-      "eval_runtime": 2.1109,
-      "eval_samples_per_second": 14.212,
-      "eval_steps_per_second": 3.79,
+      "eval_runtime": 3.1322,
+      "eval_samples_per_second": 9.578,
+      "eval_steps_per_second": 2.554,
       "num_input_tokens_seen": 24990,
       "step": 24
     },
     {
       "epoch": 0.48695652173913045,
-      "eval_loss": 0.2881294786930084,
+      "eval_loss": 0.3061341643333435,
       "eval_num_tokens": 28628.0,
-      "eval_runtime": 2.1211,
-      "eval_samples_per_second": 14.143,
-      "eval_steps_per_second": 3.772,
+      "eval_runtime": 3.1408,
+      "eval_samples_per_second": 9.552,
+      "eval_steps_per_second": 2.547,
       "num_input_tokens_seen": 28628,
       "step": 28
     },
     {
       "epoch": 0.5217391304347826,
-      "grad_norm": 6.022609233856201,
+      "grad_norm": 3.9868009090423584,
       "learning_rate": 0.00011490240573865023,
-      "loss": 0.14516949653625488,
+      "loss": 0.19884855747222902,
       "num_input_tokens_seen": 30604,
       "num_tokens": 30604.0,
       "step": 30,
-      "train_runtime": 233.5004,
-      "train_tokens_per_second": 131.066
+      "train_runtime": 304.1359,
+      "train_tokens_per_second": 100.626
     },
     {
       "epoch": 0.5565217391304348,
-      "eval_loss": 0.22245244681835175,
+      "eval_loss": 0.25844958424568176,
       "eval_num_tokens": 33017.0,
-      "eval_runtime": 2.0997,
-      "eval_samples_per_second": 14.288,
-      "eval_steps_per_second": 3.81,
+      "eval_runtime": 3.1661,
+      "eval_samples_per_second": 9.475,
+      "eval_steps_per_second": 2.527,
       "num_input_tokens_seen": 33017,
       "step": 32
     },
     {
       "epoch": 0.6260869565217392,
-      "eval_loss": 0.1709093600511551,
+      "eval_loss": 0.21466659009456635,
       "eval_num_tokens": 37556.0,
-      "eval_runtime": 2.1061,
-      "eval_samples_per_second": 14.244,
-      "eval_steps_per_second": 3.798,
+      "eval_runtime": 3.2027,
+      "eval_samples_per_second": 9.367,
+      "eval_steps_per_second": 2.498,
       "num_input_tokens_seen": 37556,
       "step": 36
     },
     {
       "epoch": 0.6956521739130435,
-      "grad_norm": 1.5943447351455688,
+      "grad_norm": 2.9145026206970215,
       "learning_rate": 7.5e-05,
-      "loss": 0.09838705658912658,
+      "loss": 0.11073940992355347,
       "num_input_tokens_seen": 43339,
       "num_tokens": 43339.0,
       "step": 40,
-      "train_runtime": 282.5787,
-      "train_tokens_per_second": 153.37
+      "train_runtime": 370.1019,
+      "train_tokens_per_second": 117.1
     },
     {
       "epoch": 0.6956521739130435,
-      "eval_loss": 0.17494265735149384,
+      "eval_loss": 0.20814543962478638,
       "eval_num_tokens": 43339.0,
-      "eval_runtime": 2.1383,
-      "eval_samples_per_second": 14.03,
-      "eval_steps_per_second": 3.741,
+      "eval_runtime": 3.2117,
+      "eval_samples_per_second": 9.341,
+      "eval_steps_per_second": 2.491,
       "num_input_tokens_seen": 43339,
       "step": 40
     },
     {
       "epoch": 0.7652173913043478,
-      "eval_loss": 0.1449701488018036,
+      "eval_loss": 0.13490967452526093,
       "eval_num_tokens": 51343.0,
-      "eval_runtime": 2.1082,
-      "eval_samples_per_second": 14.23,
-      "eval_steps_per_second": 3.795,
+      "eval_runtime": 3.1593,
+      "eval_samples_per_second": 9.496,
+      "eval_steps_per_second": 2.532,
       "num_input_tokens_seen": 51343,
       "step": 44
     },
     {
       "epoch": 0.8347826086956521,
-      "eval_loss": 0.12524840235710144,
+      "eval_loss": 0.11187396198511124,
       "eval_num_tokens": 55787.0,
-      "eval_runtime": 2.1038,
-      "eval_samples_per_second": 14.26,
-      "eval_steps_per_second": 3.803,
+      "eval_runtime": 3.1694,
+      "eval_samples_per_second": 9.465,
+      "eval_steps_per_second": 2.524,
       "num_input_tokens_seen": 55787,
       "step": 48
     },
     {
       "epoch": 0.8695652173913043,
-      "grad_norm": 1.0154601335525513,
+      "grad_norm": 1.7305588722229004,
       "learning_rate": 3.5097594261349764e-05,
-      "loss": 0.07467616200447083,
+      "loss": 0.10026404857635499,
       "num_input_tokens_seen": 56637,
       "num_tokens": 56637.0,
       "step": 50,
-      "train_runtime": 341.3231,
-      "train_tokens_per_second": 165.934
+      "train_runtime": 448.4165,
+      "train_tokens_per_second": 126.304
     },
     {
       "epoch": 0.9043478260869565,
-      "eval_loss": 0.11500633507966995,
+      "eval_loss": 0.11084464192390442,
       "eval_num_tokens": 60640.0,
-      "eval_runtime": 2.0996,
-      "eval_samples_per_second": 14.288,
-      "eval_steps_per_second": 3.81,
+      "eval_runtime": 3.1284,
+      "eval_samples_per_second": 9.589,
+      "eval_steps_per_second": 2.557,
       "num_input_tokens_seen": 60640,
       "step": 52
     },
     {
       "epoch": 0.9739130434782609,
-      "eval_loss": 0.10254310816526413,
+      "eval_loss": 0.10104832798242569,
       "eval_num_tokens": 66252.0,
-      "eval_runtime": 2.1287,
-      "eval_samples_per_second": 14.093,
-      "eval_steps_per_second": 3.758,
+      "eval_runtime": 3.1587,
+      "eval_samples_per_second": 9.498,
+      "eval_steps_per_second": 2.533,
       "num_input_tokens_seen": 66252,
       "step": 56
     },
     {
       "epoch": 1.0,
-      "eval_loss": 0.09790374338626862,
+      "eval_loss": 0.09292192757129669,
       "eval_num_tokens": 67398.0,
-      "eval_runtime": 2.0904,
-      "eval_samples_per_second": 14.351,
-      "eval_steps_per_second": 3.827,
+      "eval_runtime": 3.1225,
+      "eval_samples_per_second": 9.608,
+      "eval_steps_per_second": 2.562,
       "num_input_tokens_seen": 67398,
       "step": 58
     }
@@ -243,7 +243,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 270336498597888.0,
+  "total_flos": 373164151930880.0,
   "train_batch_size": 1,
   "trial_name": null,
   "trial_params": null