SFT checkpoint step 58

checkpoints/checkpoint-58/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ea6862c6cf353de778a159ea0f907cb011f50089bf659d42648539746d0937b9
+oid sha256:db17a50472276b863efebaae63dd5173d1d71e5afe697186713de94b3e671fd6
 size 204500912

checkpoints/checkpoint-58/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:df7502b47f9417a68d9d52633285620465862a5ef2bedc8286d9886dbf7e6990
+oid sha256:a947c02b62f310e91e205429a144af3de2f7ccf8b8f6ebe7175d109091126c1b
 size 620569547

checkpoints/checkpoint-58/scaler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:94daf24225993d49e3b9dae55d8f50c9ca8d34c6de2f05ec79c39db3cde15c72
+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": 514.7618689537048, "global_step": 58}
+{"consumed_seconds": 535.0243752002716, "global_step": 58}

checkpoints/checkpoint-58/trainer_state.json

@@ -17,15 +17,15 @@
       "num_input_tokens_seen": 1494,
       "num_tokens": 1494.0,
       "step": 1,
-      "train_runtime": 44.9864,
-      "train_tokens_per_second": 33.21
+      "train_runtime": 42.4619,
+      "train_tokens_per_second": 35.185
     },
     {
       "epoch": 0.06956521739130435,
       "eval_loss": 5.31978178024292,
       "eval_num_tokens": 4980.0,
-      "eval_runtime": 28.6663,
-      "eval_samples_per_second": 1.047,
+      "eval_runtime": 28.7144,
+      "eval_samples_per_second": 1.045,
       "eval_steps_per_second": 0.279,
       "num_input_tokens_seen": 4980,
       "step": 4
@@ -34,194 +34,194 @@
       "epoch": 0.1391304347826087,
       "eval_loss": 5.31978178024292,
       "eval_num_tokens": 9004.0,
-      "eval_runtime": 3.5559,
-      "eval_samples_per_second": 8.437,
-      "eval_steps_per_second": 2.25,
+      "eval_runtime": 3.4006,
+      "eval_samples_per_second": 8.822,
+      "eval_steps_per_second": 2.353,
       "num_input_tokens_seen": 9004,
       "step": 8
     },
     {
       "epoch": 0.17391304347826086,
-      "grad_norm": 28.138591766357422,
+      "grad_norm": 28.1390438079834,
       "learning_rate": 0.0,
       "loss": 2.5477417839898004,
       "num_input_tokens_seen": 11145,
       "num_tokens": 11145.0,
       "step": 10,
-      "train_runtime": 145.0656,
-      "train_tokens_per_second": 76.827
+      "train_runtime": 146.4448,
+      "train_tokens_per_second": 76.104
     },
     {
       "epoch": 0.20869565217391303,
-      "eval_loss": 1.795594573020935,
+      "eval_loss": 1.7850792407989502,
       "eval_num_tokens": 13293.0,
-      "eval_runtime": 3.5484,
-      "eval_samples_per_second": 8.455,
-      "eval_steps_per_second": 2.255,
+      "eval_runtime": 3.4816,
+      "eval_samples_per_second": 8.617,
+      "eval_steps_per_second": 2.298,
       "num_input_tokens_seen": 13293,
       "step": 12
     },
     {
       "epoch": 0.2782608695652174,
-      "eval_loss": 0.5860968232154846,
+      "eval_loss": 0.5965744256973267,
       "eval_num_tokens": 17742.0,
-      "eval_runtime": 3.4658,
-      "eval_samples_per_second": 8.656,
-      "eval_steps_per_second": 2.308,
+      "eval_runtime": 3.4391,
+      "eval_samples_per_second": 8.723,
+      "eval_steps_per_second": 2.326,
       "num_input_tokens_seen": 17742,
       "step": 16
     },
     {
       "epoch": 0.34782608695652173,
-      "grad_norm": 7.353201389312744,
-      "learning_rate": 0.0001442909649383465,
-      "loss": 0.8008382797241211,
+      "grad_norm": 6.518080711364746,
+      "learning_rate": 0.0001425726650926814,
+      "loss": 0.7919160842895507,
       "num_input_tokens_seen": 20264,
       "num_tokens": 20264.0,
       "step": 20,
-      "train_runtime": 227.664,
-      "train_tokens_per_second": 89.008
+      "train_runtime": 241.9804,
+      "train_tokens_per_second": 83.742
     },
     {
       "epoch": 0.34782608695652173,
-      "eval_loss": 0.39688172936439514,
+      "eval_loss": 0.6307523250579834,
       "eval_num_tokens": 20264.0,
-      "eval_runtime": 3.494,
-      "eval_samples_per_second": 8.586,
-      "eval_steps_per_second": 2.29,
+      "eval_runtime": 3.512,
+      "eval_samples_per_second": 8.542,
+      "eval_steps_per_second": 2.278,
       "num_input_tokens_seen": 20264,
       "step": 20
     },
     {
       "epoch": 0.41739130434782606,
-      "eval_loss": 0.5452550053596497,
+      "eval_loss": 0.3135659396648407,
       "eval_num_tokens": 24990.0,
-      "eval_runtime": 3.5856,
-      "eval_samples_per_second": 8.367,
-      "eval_steps_per_second": 2.231,
+      "eval_runtime": 3.4341,
+      "eval_samples_per_second": 8.736,
+      "eval_steps_per_second": 2.33,
       "num_input_tokens_seen": 24990,
       "step": 24
     },
     {
       "epoch": 0.48695652173913045,
-      "eval_loss": 0.31608301401138306,
+      "eval_loss": 0.4761301875114441,
       "eval_num_tokens": 28628.0,
-      "eval_runtime": 3.5529,
-      "eval_samples_per_second": 8.444,
-      "eval_steps_per_second": 2.252,
+      "eval_runtime": 3.4198,
+      "eval_samples_per_second": 8.772,
+      "eval_steps_per_second": 2.339,
       "num_input_tokens_seen": 28628,
       "step": 28
     },
     {
       "epoch": 0.5217391304347826,
-      "grad_norm": 2.936347007751465,
-      "learning_rate": 0.00011840034721348544,
-      "loss": 0.20213119983673095,
+      "grad_norm": 3.650630235671997,
+      "learning_rate": 0.00011490240573865023,
+      "loss": 0.22239422798156738,
       "num_input_tokens_seen": 30604,
       "num_tokens": 30604.0,
       "step": 30,
-      "train_runtime": 314.4011,
-      "train_tokens_per_second": 97.341
+      "train_runtime": 326.8185,
+      "train_tokens_per_second": 93.642
     },
     {
       "epoch": 0.5565217391304348,
-      "eval_loss": 0.3312884569168091,
+      "eval_loss": 0.2599254846572876,
       "eval_num_tokens": 33017.0,
-      "eval_runtime": 3.5533,
-      "eval_samples_per_second": 8.443,
-      "eval_steps_per_second": 2.251,
+      "eval_runtime": 3.409,
+      "eval_samples_per_second": 8.8,
+      "eval_steps_per_second": 2.347,
       "num_input_tokens_seen": 33017,
       "step": 32
     },
     {
       "epoch": 0.6260869565217392,
-      "eval_loss": 0.23424790799617767,
+      "eval_loss": 0.17896154522895813,
       "eval_num_tokens": 37556.0,
-      "eval_runtime": 3.4971,
-      "eval_samples_per_second": 8.579,
-      "eval_steps_per_second": 2.288,
+      "eval_runtime": 3.3953,
+      "eval_samples_per_second": 8.836,
+      "eval_steps_per_second": 2.356,
       "num_input_tokens_seen": 37556,
       "step": 36
     },
     {
       "epoch": 0.6956521739130435,
-      "grad_norm": 2.543781280517578,
-      "learning_rate": 7.920528354278937e-05,
-      "loss": 0.12761373519897462,
+      "grad_norm": 3.054882049560547,
+      "learning_rate": 7.5e-05,
+      "loss": 0.1124645471572876,
       "num_input_tokens_seen": 43339,
       "num_tokens": 43339.0,
       "step": 40,
-      "train_runtime": 381.0466,
-      "train_tokens_per_second": 113.737
+      "train_runtime": 400.6098,
+      "train_tokens_per_second": 108.183
     },
     {
       "epoch": 0.6956521739130435,
-      "eval_loss": 0.20891477167606354,
+      "eval_loss": 0.19172506034374237,
       "eval_num_tokens": 43339.0,
-      "eval_runtime": 3.4787,
-      "eval_samples_per_second": 8.624,
-      "eval_steps_per_second": 2.3,
+      "eval_runtime": 3.3581,
+      "eval_samples_per_second": 8.934,
+      "eval_steps_per_second": 2.382,
       "num_input_tokens_seen": 43339,
       "step": 40
     },
     {
       "epoch": 0.7652173913043478,
-      "eval_loss": 0.16005884110927582,
+      "eval_loss": 0.14692261815071106,
       "eval_num_tokens": 51343.0,
-      "eval_runtime": 3.5211,
-      "eval_samples_per_second": 8.52,
-      "eval_steps_per_second": 2.272,
+      "eval_runtime": 3.409,
+      "eval_samples_per_second": 8.8,
+      "eval_steps_per_second": 2.347,
       "num_input_tokens_seen": 51343,
       "step": 44
     },
     {
       "epoch": 0.8347826086956521,
-      "eval_loss": 0.1316540390253067,
+      "eval_loss": 0.13185091316699982,
       "eval_num_tokens": 55787.0,
-      "eval_runtime": 3.5293,
-      "eval_samples_per_second": 8.5,
-      "eval_steps_per_second": 2.267,
+      "eval_runtime": 3.4735,
+      "eval_samples_per_second": 8.637,
+      "eval_steps_per_second": 2.303,
       "num_input_tokens_seen": 55787,
       "step": 48
     },
     {
       "epoch": 0.8695652173913043,
-      "grad_norm": 2.2746105194091797,
-      "learning_rate": 3.872108346710701e-05,
-      "loss": 0.103130304813385,
+      "grad_norm": 2.5227229595184326,
+      "learning_rate": 3.5097594261349764e-05,
+      "loss": 0.08426350355148315,
       "num_input_tokens_seen": 56637,
       "num_tokens": 56637.0,
       "step": 50,
-      "train_runtime": 459.3749,
-      "train_tokens_per_second": 123.291
+      "train_runtime": 479.1478,
+      "train_tokens_per_second": 118.204
     },
     {
       "epoch": 0.9043478260869565,
-      "eval_loss": 0.10423395037651062,
+      "eval_loss": 0.09836789965629578,
       "eval_num_tokens": 60640.0,
-      "eval_runtime": 3.4932,
-      "eval_samples_per_second": 8.588,
-      "eval_steps_per_second": 2.29,
+      "eval_runtime": 3.4114,
+      "eval_samples_per_second": 8.794,
+      "eval_steps_per_second": 2.345,
       "num_input_tokens_seen": 60640,
       "step": 52
     },
     {
       "epoch": 0.9739130434782609,
-      "eval_loss": 0.08810122311115265,
+      "eval_loss": 0.07989717274904251,
       "eval_num_tokens": 66252.0,
-      "eval_runtime": 3.5537,
-      "eval_samples_per_second": 8.442,
-      "eval_steps_per_second": 2.251,
+      "eval_runtime": 3.4003,
+      "eval_samples_per_second": 8.823,
+      "eval_steps_per_second": 2.353,
       "num_input_tokens_seen": 66252,
       "step": 56
     },
     {
       "epoch": 1.0,
-      "eval_loss": 0.0836324393749237,
+      "eval_loss": 0.0754588395357132,
       "eval_num_tokens": 67398.0,
-      "eval_runtime": 3.4688,
-      "eval_samples_per_second": 8.649,
-      "eval_steps_per_second": 2.306,
+      "eval_runtime": 3.3898,
+      "eval_samples_per_second": 8.85,
+      "eval_steps_per_second": 2.36,
       "num_input_tokens_seen": 67398,
       "step": 58
     }