Model save
Browse files- all_results.json +5 -5
- train_results.json +5 -5
- trainer_state.json +785 -25
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"total_flos":
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 15921,
|
| 6 |
-
"train_samples_per_second": 0.
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"total_flos": 2.496956368699785e+18,
|
| 3 |
+
"train_loss": 0.33275703743100166,
|
| 4 |
+
"train_runtime": 21613.1094,
|
| 5 |
"train_samples": 15921,
|
| 6 |
+
"train_samples_per_second": 0.863,
|
| 7 |
+
"train_steps_per_second": 0.005
|
| 8 |
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"total_flos":
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 15921,
|
| 6 |
-
"train_samples_per_second": 0.
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"total_flos": 2.496956368699785e+18,
|
| 3 |
+
"train_loss": 0.33275703743100166,
|
| 4 |
+
"train_runtime": 21613.1094,
|
| 5 |
"train_samples": 15921,
|
| 6 |
+
"train_samples_per_second": 0.863,
|
| 7 |
+
"train_steps_per_second": 0.005
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -2,16 +2,16 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 0
|
| 6 |
"eval_steps": 500,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.05128205128205128,
|
| 14 |
-
"grad_norm": 2.
|
| 15 |
"learning_rate": 0.0,
|
| 16 |
"loss": 0.5757,
|
| 17 |
"num_tokens": 6240601.0,
|
|
@@ -19,51 +19,811 @@
|
|
| 19 |
},
|
| 20 |
{
|
| 21 |
"epoch": 0.10256410256410256,
|
| 22 |
-
"grad_norm": 2.
|
| 23 |
-
"learning_rate":
|
| 24 |
"loss": 0.5766,
|
| 25 |
"num_tokens": 12512622.0,
|
| 26 |
"step": 2
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 0.15384615384615385,
|
| 30 |
-
"grad_norm":
|
| 31 |
-
"learning_rate":
|
| 32 |
-
"loss": 0.
|
| 33 |
"num_tokens": 18771271.0,
|
| 34 |
"step": 3
|
| 35 |
},
|
| 36 |
{
|
| 37 |
"epoch": 0.20512820512820512,
|
| 38 |
-
"grad_norm":
|
| 39 |
-
"learning_rate":
|
| 40 |
-
"loss": 0.
|
| 41 |
"num_tokens": 25037339.0,
|
| 42 |
"step": 4
|
| 43 |
},
|
| 44 |
{
|
| 45 |
"epoch": 0.2564102564102564,
|
| 46 |
-
"grad_norm":
|
| 47 |
-
"learning_rate":
|
| 48 |
-
"loss": 0.
|
| 49 |
"num_tokens": 31263584.0,
|
| 50 |
"step": 5
|
| 51 |
},
|
| 52 |
{
|
| 53 |
-
"epoch": 0.
|
| 54 |
-
"
|
| 55 |
-
"
|
| 56 |
-
"
|
| 57 |
-
"
|
| 58 |
-
"
|
| 59 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 60 |
}
|
| 61 |
],
|
| 62 |
"logging_steps": 1,
|
| 63 |
-
"max_steps":
|
| 64 |
"num_input_tokens_seen": 0,
|
| 65 |
-
"num_train_epochs":
|
| 66 |
-
"save_steps":
|
| 67 |
"stateful_callbacks": {
|
| 68 |
"TrainerControl": {
|
| 69 |
"args": {
|
|
@@ -76,7 +836,7 @@
|
|
| 76 |
"attributes": {}
|
| 77 |
}
|
| 78 |
},
|
| 79 |
-
"total_flos":
|
| 80 |
"train_batch_size": 1,
|
| 81 |
"trial_name": null,
|
| 82 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 5.0,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 100,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.05128205128205128,
|
| 14 |
+
"grad_norm": 2.718615561225469,
|
| 15 |
"learning_rate": 0.0,
|
| 16 |
"loss": 0.5757,
|
| 17 |
"num_tokens": 6240601.0,
|
|
|
|
| 19 |
},
|
| 20 |
{
|
| 21 |
"epoch": 0.10256410256410256,
|
| 22 |
+
"grad_norm": 2.7384427318262325,
|
| 23 |
+
"learning_rate": 1.3333333333333333e-05,
|
| 24 |
"loss": 0.5766,
|
| 25 |
"num_tokens": 12512622.0,
|
| 26 |
"step": 2
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 0.15384615384615385,
|
| 30 |
+
"grad_norm": 1.9843963750520588,
|
| 31 |
+
"learning_rate": 2.6666666666666667e-05,
|
| 32 |
+
"loss": 0.5242,
|
| 33 |
"num_tokens": 18771271.0,
|
| 34 |
"step": 3
|
| 35 |
},
|
| 36 |
{
|
| 37 |
"epoch": 0.20512820512820512,
|
| 38 |
+
"grad_norm": 2.12777357229112,
|
| 39 |
+
"learning_rate": 4e-05,
|
| 40 |
+
"loss": 0.4917,
|
| 41 |
"num_tokens": 25037339.0,
|
| 42 |
"step": 4
|
| 43 |
},
|
| 44 |
{
|
| 45 |
"epoch": 0.2564102564102564,
|
| 46 |
+
"grad_norm": 2.062032488995502,
|
| 47 |
+
"learning_rate": 3.9990560242819274e-05,
|
| 48 |
+
"loss": 0.4871,
|
| 49 |
"num_tokens": 31263584.0,
|
| 50 |
"step": 5
|
| 51 |
},
|
| 52 |
{
|
| 53 |
+
"epoch": 0.3076923076923077,
|
| 54 |
+
"grad_norm": 0.7945484310641507,
|
| 55 |
+
"learning_rate": 3.996225087227881e-05,
|
| 56 |
+
"loss": 0.4538,
|
| 57 |
+
"num_tokens": 37544028.0,
|
| 58 |
+
"step": 6
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.358974358974359,
|
| 62 |
+
"grad_norm": 0.860696592609233,
|
| 63 |
+
"learning_rate": 3.991510158099905e-05,
|
| 64 |
+
"loss": 0.4585,
|
| 65 |
+
"num_tokens": 43782393.0,
|
| 66 |
+
"step": 7
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.41025641025641024,
|
| 70 |
+
"grad_norm": 0.505429642395346,
|
| 71 |
+
"learning_rate": 3.9849161822075655e-05,
|
| 72 |
+
"loss": 0.4423,
|
| 73 |
+
"num_tokens": 50063870.0,
|
| 74 |
+
"step": 8
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.46153846153846156,
|
| 78 |
+
"grad_norm": 0.7741144627099796,
|
| 79 |
+
"learning_rate": 3.976450075721003e-05,
|
| 80 |
+
"loss": 0.4439,
|
| 81 |
+
"num_tokens": 56308984.0,
|
| 82 |
+
"step": 9
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.5128205128205128,
|
| 86 |
+
"grad_norm": 0.40862829974318693,
|
| 87 |
+
"learning_rate": 3.9661207184168305e-05,
|
| 88 |
+
"loss": 0.4234,
|
| 89 |
+
"num_tokens": 62545986.0,
|
| 90 |
+
"step": 10
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.5641025641025641,
|
| 94 |
+
"grad_norm": 0.4739349747902679,
|
| 95 |
+
"learning_rate": 3.953938944364467e-05,
|
| 96 |
+
"loss": 0.4171,
|
| 97 |
+
"num_tokens": 68774917.0,
|
| 98 |
+
"step": 11
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.6153846153846154,
|
| 102 |
+
"grad_norm": 0.36916258563551035,
|
| 103 |
+
"learning_rate": 3.939917530562701e-05,
|
| 104 |
+
"loss": 0.4182,
|
| 105 |
+
"num_tokens": 75056519.0,
|
| 106 |
+
"step": 12
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 0.6666666666666666,
|
| 110 |
+
"grad_norm": 0.33745730043472655,
|
| 111 |
+
"learning_rate": 3.9240711835383766e-05,
|
| 112 |
+
"loss": 0.3968,
|
| 113 |
+
"num_tokens": 81319039.0,
|
| 114 |
+
"step": 13
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.717948717948718,
|
| 118 |
+
"grad_norm": 0.3682996169134089,
|
| 119 |
+
"learning_rate": 3.9064165239212874e-05,
|
| 120 |
+
"loss": 0.4167,
|
| 121 |
+
"num_tokens": 87572805.0,
|
| 122 |
+
"step": 14
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.7692307692307693,
|
| 126 |
+
"grad_norm": 0.2610361232979272,
|
| 127 |
+
"learning_rate": 3.886972069011419e-05,
|
| 128 |
+
"loss": 0.4076,
|
| 129 |
+
"num_tokens": 93846455.0,
|
| 130 |
+
"step": 15
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.8205128205128205,
|
| 134 |
+
"grad_norm": 0.2574923493976134,
|
| 135 |
+
"learning_rate": 3.865758213356868e-05,
|
| 136 |
+
"loss": 0.3931,
|
| 137 |
+
"num_tokens": 100128858.0,
|
| 138 |
+
"step": 16
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.8717948717948718,
|
| 142 |
+
"grad_norm": 0.25582043047449654,
|
| 143 |
+
"learning_rate": 3.8427972073627724e-05,
|
| 144 |
+
"loss": 0.4034,
|
| 145 |
+
"num_tokens": 106382537.0,
|
| 146 |
+
"step": 17
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.9230769230769231,
|
| 150 |
+
"grad_norm": 0.22629074274698618,
|
| 151 |
+
"learning_rate": 3.818113133953712e-05,
|
| 152 |
+
"loss": 0.3938,
|
| 153 |
+
"num_tokens": 112641128.0,
|
| 154 |
+
"step": 18
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.9743589743589743,
|
| 158 |
+
"grad_norm": 0.1995507432503149,
|
| 159 |
+
"learning_rate": 3.791731883314043e-05,
|
| 160 |
+
"loss": 0.4007,
|
| 161 |
+
"num_tokens": 118920817.0,
|
| 162 |
+
"step": 19
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 1.0,
|
| 166 |
+
"grad_norm": 0.1995507432503149,
|
| 167 |
+
"learning_rate": 3.763681125732672e-05,
|
| 168 |
+
"loss": 0.3794,
|
| 169 |
+
"num_tokens": 120361037.0,
|
| 170 |
+
"step": 20
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 1.0512820512820513,
|
| 174 |
+
"grad_norm": 0.34048667122818044,
|
| 175 |
+
"learning_rate": 3.733990282580745e-05,
|
| 176 |
+
"loss": 0.3631,
|
| 177 |
+
"num_tokens": 126640793.0,
|
| 178 |
+
"step": 21
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 1.1025641025641026,
|
| 182 |
+
"grad_norm": 0.19615438837401894,
|
| 183 |
+
"learning_rate": 3.7026904954526884e-05,
|
| 184 |
+
"loss": 0.3736,
|
| 185 |
+
"num_tokens": 132891349.0,
|
| 186 |
+
"step": 22
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.1538461538461537,
|
| 190 |
+
"grad_norm": 0.16644896220477234,
|
| 191 |
+
"learning_rate": 3.6698145935029794e-05,
|
| 192 |
+
"loss": 0.3773,
|
| 193 |
+
"num_tokens": 139140959.0,
|
| 194 |
+
"step": 23
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 1.205128205128205,
|
| 198 |
+
"grad_norm": 0.22809017244958504,
|
| 199 |
+
"learning_rate": 3.6353970590128975e-05,
|
| 200 |
+
"loss": 0.3763,
|
| 201 |
+
"num_tokens": 145391352.0,
|
| 202 |
+
"step": 24
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 1.2564102564102564,
|
| 206 |
+
"grad_norm": 0.16923970327686066,
|
| 207 |
+
"learning_rate": 3.599473991223369e-05,
|
| 208 |
+
"loss": 0.3644,
|
| 209 |
+
"num_tokens": 151670841.0,
|
| 210 |
+
"step": 25
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 1.3076923076923077,
|
| 214 |
+
"grad_norm": 0.17762206878485046,
|
| 215 |
+
"learning_rate": 3.5620830684718515e-05,
|
| 216 |
+
"loss": 0.3614,
|
| 217 |
+
"num_tokens": 157951707.0,
|
| 218 |
+
"step": 26
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 1.358974358974359,
|
| 222 |
+
"grad_norm": 0.17590149731091576,
|
| 223 |
+
"learning_rate": 3.523263508672961e-05,
|
| 224 |
+
"loss": 0.3678,
|
| 225 |
+
"num_tokens": 164210738.0,
|
| 226 |
+
"step": 27
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 1.4102564102564101,
|
| 230 |
+
"grad_norm": 0.1385680492384403,
|
| 231 |
+
"learning_rate": 3.483056028184293e-05,
|
| 232 |
+
"loss": 0.3636,
|
| 233 |
+
"num_tokens": 170465186.0,
|
| 234 |
+
"step": 28
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 1.4615384615384617,
|
| 238 |
+
"grad_norm": 0.15559805698049264,
|
| 239 |
+
"learning_rate": 3.441502799100588e-05,
|
| 240 |
+
"loss": 0.3728,
|
| 241 |
+
"num_tokens": 176747206.0,
|
| 242 |
+
"step": 29
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.5128205128205128,
|
| 246 |
+
"grad_norm": 0.14563346219653364,
|
| 247 |
+
"learning_rate": 3.398647405021026e-05,
|
| 248 |
+
"loss": 0.3653,
|
| 249 |
+
"num_tokens": 183011405.0,
|
| 250 |
+
"step": 30
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 1.564102564102564,
|
| 254 |
+
"grad_norm": 0.13958012967685854,
|
| 255 |
+
"learning_rate": 3.354534795336052e-05,
|
| 256 |
+
"loss": 0.3701,
|
| 257 |
+
"num_tokens": 189245121.0,
|
| 258 |
+
"step": 31
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 1.6153846153846154,
|
| 262 |
+
"grad_norm": 0.11987296091737516,
|
| 263 |
+
"learning_rate": 3.3092112380816696e-05,
|
| 264 |
+
"loss": 0.3543,
|
| 265 |
+
"num_tokens": 195525140.0,
|
| 266 |
+
"step": 32
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 1.6666666666666665,
|
| 270 |
+
"grad_norm": 0.14628950900383383,
|
| 271 |
+
"learning_rate": 3.262724271410661e-05,
|
| 272 |
+
"loss": 0.3564,
|
| 273 |
+
"num_tokens": 201755990.0,
|
| 274 |
+
"step": 33
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 1.717948717948718,
|
| 278 |
+
"grad_norm": 0.10624435308415683,
|
| 279 |
+
"learning_rate": 3.2151226537316315e-05,
|
| 280 |
+
"loss": 0.3555,
|
| 281 |
+
"num_tokens": 208001091.0,
|
| 282 |
+
"step": 34
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 1.7692307692307692,
|
| 286 |
+
"grad_norm": 0.13312768270680841,
|
| 287 |
+
"learning_rate": 3.166456312568171e-05,
|
| 288 |
+
"loss": 0.367,
|
| 289 |
+
"num_tokens": 214269555.0,
|
| 290 |
+
"step": 35
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 1.8205128205128205,
|
| 294 |
+
"grad_norm": 0.11882236522455852,
|
| 295 |
+
"learning_rate": 3.116776292191774e-05,
|
| 296 |
+
"loss": 0.3492,
|
| 297 |
+
"num_tokens": 220504752.0,
|
| 298 |
+
"step": 36
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 1.8717948717948718,
|
| 302 |
+
"grad_norm": 0.1108379812394889,
|
| 303 |
+
"learning_rate": 3.0661347000834496e-05,
|
| 304 |
+
"loss": 0.3603,
|
| 305 |
+
"num_tokens": 226746431.0,
|
| 306 |
+
"step": 37
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 1.9230769230769231,
|
| 310 |
+
"grad_norm": 0.1209856619633538,
|
| 311 |
+
"learning_rate": 3.0145846522801703e-05,
|
| 312 |
+
"loss": 0.364,
|
| 313 |
+
"num_tokens": 233000985.0,
|
| 314 |
+
"step": 38
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 1.9743589743589745,
|
| 318 |
+
"grad_norm": 0.11419043533635012,
|
| 319 |
+
"learning_rate": 2.962180217663483e-05,
|
| 320 |
+
"loss": 0.3555,
|
| 321 |
+
"num_tokens": 239282562.0,
|
| 322 |
+
"step": 39
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 2.0,
|
| 326 |
+
"grad_norm": 0.18167550021785242,
|
| 327 |
+
"learning_rate": 2.908976361248717e-05,
|
| 328 |
+
"loss": 0.3489,
|
| 329 |
+
"num_tokens": 240721535.0,
|
| 330 |
+
"step": 40
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 2.051282051282051,
|
| 334 |
+
"grad_norm": 0.15212860008317705,
|
| 335 |
+
"learning_rate": 2.855028886534278e-05,
|
| 336 |
+
"loss": 0.3218,
|
| 337 |
+
"num_tokens": 246968297.0,
|
| 338 |
+
"step": 41
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 2.1025641025641026,
|
| 342 |
+
"grad_norm": 0.1298804530561056,
|
| 343 |
+
"learning_rate": 2.8003943769714776e-05,
|
| 344 |
+
"loss": 0.3299,
|
| 345 |
+
"num_tokens": 253234954.0,
|
| 346 |
+
"step": 42
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 2.1538461538461537,
|
| 350 |
+
"grad_norm": 0.1405980356910723,
|
| 351 |
+
"learning_rate": 2.7451301366163116e-05,
|
| 352 |
+
"loss": 0.3196,
|
| 353 |
+
"num_tokens": 259516218.0,
|
| 354 |
+
"step": 43
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 2.2051282051282053,
|
| 358 |
+
"grad_norm": 0.10364248867045861,
|
| 359 |
+
"learning_rate": 2.6892941300254176e-05,
|
| 360 |
+
"loss": 0.3287,
|
| 361 |
+
"num_tokens": 265777892.0,
|
| 362 |
+
"step": 44
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 2.2564102564102564,
|
| 366 |
+
"grad_norm": 0.1283238256584598,
|
| 367 |
+
"learning_rate": 2.6329449214592568e-05,
|
| 368 |
+
"loss": 0.3268,
|
| 369 |
+
"num_tokens": 272038252.0,
|
| 370 |
+
"step": 45
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.3076923076923075,
|
| 374 |
+
"grad_norm": 0.09763592234171578,
|
| 375 |
+
"learning_rate": 2.5761416134562955e-05,
|
| 376 |
+
"loss": 0.3126,
|
| 377 |
+
"num_tokens": 278296569.0,
|
| 378 |
+
"step": 46
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 2.358974358974359,
|
| 382 |
+
"grad_norm": 0.11539957768743261,
|
| 383 |
+
"learning_rate": 2.5189437848426016e-05,
|
| 384 |
+
"loss": 0.3241,
|
| 385 |
+
"num_tokens": 284513318.0,
|
| 386 |
+
"step": 47
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 2.41025641025641,
|
| 390 |
+
"grad_norm": 0.10255646114610974,
|
| 391 |
+
"learning_rate": 2.461411428241883e-05,
|
| 392 |
+
"loss": 0.313,
|
| 393 |
+
"num_tokens": 290793851.0,
|
| 394 |
+
"step": 48
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 2.4615384615384617,
|
| 398 |
+
"grad_norm": 0.11056138494167911,
|
| 399 |
+
"learning_rate": 2.403604887151512e-05,
|
| 400 |
+
"loss": 0.3142,
|
| 401 |
+
"num_tokens": 297038281.0,
|
| 402 |
+
"step": 49
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 2.5128205128205128,
|
| 406 |
+
"grad_norm": 0.09087426235660283,
|
| 407 |
+
"learning_rate": 2.3455847926505283e-05,
|
| 408 |
+
"loss": 0.3237,
|
| 409 |
+
"num_tokens": 303309073.0,
|
| 410 |
+
"step": 50
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 2.564102564102564,
|
| 414 |
+
"grad_norm": 0.09942623723641979,
|
| 415 |
+
"learning_rate": 2.287411999806007e-05,
|
| 416 |
+
"loss": 0.31,
|
| 417 |
+
"num_tokens": 309590829.0,
|
| 418 |
+
"step": 51
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 2.6153846153846154,
|
| 422 |
+
"grad_norm": 0.0814248146482854,
|
| 423 |
+
"learning_rate": 2.2291475238445033e-05,
|
| 424 |
+
"loss": 0.3211,
|
| 425 |
+
"num_tokens": 315858116.0,
|
| 426 |
+
"step": 52
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 2.6666666666666665,
|
| 430 |
+
"grad_norm": 0.09568614448485548,
|
| 431 |
+
"learning_rate": 2.1708524761554973e-05,
|
| 432 |
+
"loss": 0.3177,
|
| 433 |
+
"num_tokens": 322118809.0,
|
| 434 |
+
"step": 53
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 2.717948717948718,
|
| 438 |
+
"grad_norm": 0.07964923658403533,
|
| 439 |
+
"learning_rate": 2.112588000193994e-05,
|
| 440 |
+
"loss": 0.319,
|
| 441 |
+
"num_tokens": 328400995.0,
|
| 442 |
+
"step": 54
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 2.769230769230769,
|
| 446 |
+
"grad_norm": 0.08309011179734216,
|
| 447 |
+
"learning_rate": 2.054415207349473e-05,
|
| 448 |
+
"loss": 0.3203,
|
| 449 |
+
"num_tokens": 334639597.0,
|
| 450 |
+
"step": 55
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 2.8205128205128203,
|
| 454 |
+
"grad_norm": 0.08196424816532498,
|
| 455 |
+
"learning_rate": 1.9963951128484886e-05,
|
| 456 |
+
"loss": 0.3221,
|
| 457 |
+
"num_tokens": 340841925.0,
|
| 458 |
+
"step": 56
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 2.871794871794872,
|
| 462 |
+
"grad_norm": 0.07847272582125499,
|
| 463 |
+
"learning_rate": 1.9385885717581182e-05,
|
| 464 |
+
"loss": 0.3133,
|
| 465 |
+
"num_tokens": 347123519.0,
|
| 466 |
+
"step": 57
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 2.9230769230769234,
|
| 470 |
+
"grad_norm": 0.07759377817882507,
|
| 471 |
+
"learning_rate": 1.8810562151573993e-05,
|
| 472 |
+
"loss": 0.3039,
|
| 473 |
+
"num_tokens": 353396517.0,
|
| 474 |
+
"step": 58
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 2.9743589743589745,
|
| 478 |
+
"grad_norm": 0.08098128127962194,
|
| 479 |
+
"learning_rate": 1.823858386543705e-05,
|
| 480 |
+
"loss": 0.3097,
|
| 481 |
+
"num_tokens": 359678782.0,
|
| 482 |
+
"step": 59
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 3.0,
|
| 486 |
+
"grad_norm": 0.08098128127962194,
|
| 487 |
+
"learning_rate": 1.7670550785407444e-05,
|
| 488 |
+
"loss": 0.311,
|
| 489 |
+
"num_tokens": 361081292.0,
|
| 490 |
+
"step": 60
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 3.051282051282051,
|
| 494 |
+
"grad_norm": 0.1219526450192946,
|
| 495 |
+
"learning_rate": 1.710705869974583e-05,
|
| 496 |
+
"loss": 0.2909,
|
| 497 |
+
"num_tokens": 367363606.0,
|
| 498 |
+
"step": 61
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 3.1025641025641026,
|
| 502 |
+
"grad_norm": 0.09709390348755251,
|
| 503 |
+
"learning_rate": 1.6548698633836893e-05,
|
| 504 |
+
"loss": 0.2929,
|
| 505 |
+
"num_tokens": 373623626.0,
|
| 506 |
+
"step": 62
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 3.1538461538461537,
|
| 510 |
+
"grad_norm": 0.09032197466642125,
|
| 511 |
+
"learning_rate": 1.5996056230285237e-05,
|
| 512 |
+
"loss": 0.2881,
|
| 513 |
+
"num_tokens": 379905390.0,
|
| 514 |
+
"step": 63
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 3.2051282051282053,
|
| 518 |
+
"grad_norm": 0.09593201487472482,
|
| 519 |
+
"learning_rate": 1.5449711134657224e-05,
|
| 520 |
+
"loss": 0.2849,
|
| 521 |
+
"num_tokens": 386151857.0,
|
| 522 |
+
"step": 64
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 3.2564102564102564,
|
| 526 |
+
"grad_norm": 0.08482876184821657,
|
| 527 |
+
"learning_rate": 1.4910236387512837e-05,
|
| 528 |
+
"loss": 0.2849,
|
| 529 |
+
"num_tokens": 392391728.0,
|
| 530 |
+
"step": 65
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 3.3076923076923075,
|
| 534 |
+
"grad_norm": 0.08786917404760929,
|
| 535 |
+
"learning_rate": 1.4378197823365186e-05,
|
| 536 |
+
"loss": 0.2919,
|
| 537 |
+
"num_tokens": 398635444.0,
|
| 538 |
+
"step": 66
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 3.358974358974359,
|
| 542 |
+
"grad_norm": 0.09124789781748067,
|
| 543 |
+
"learning_rate": 1.3854153477198305e-05,
|
| 544 |
+
"loss": 0.2877,
|
| 545 |
+
"num_tokens": 404884484.0,
|
| 546 |
+
"step": 67
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 3.41025641025641,
|
| 550 |
+
"grad_norm": 0.08376420028878678,
|
| 551 |
+
"learning_rate": 1.3338652999165511e-05,
|
| 552 |
+
"loss": 0.2787,
|
| 553 |
+
"num_tokens": 411165857.0,
|
| 554 |
+
"step": 68
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 3.4615384615384617,
|
| 558 |
+
"grad_norm": 0.08461395911230184,
|
| 559 |
+
"learning_rate": 1.2832237078082272e-05,
|
| 560 |
+
"loss": 0.2874,
|
| 561 |
+
"num_tokens": 417447084.0,
|
| 562 |
+
"step": 69
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 3.5128205128205128,
|
| 566 |
+
"grad_norm": 0.0805797686999939,
|
| 567 |
+
"learning_rate": 1.2335436874318293e-05,
|
| 568 |
+
"loss": 0.2868,
|
| 569 |
+
"num_tokens": 423708369.0,
|
| 570 |
+
"step": 70
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 3.564102564102564,
|
| 574 |
+
"grad_norm": 0.0812369216368886,
|
| 575 |
+
"learning_rate": 1.1848773462683684e-05,
|
| 576 |
+
"loss": 0.2731,
|
| 577 |
+
"num_tokens": 429917854.0,
|
| 578 |
+
"step": 71
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 3.6153846153846154,
|
| 582 |
+
"grad_norm": 0.08068456709857151,
|
| 583 |
+
"learning_rate": 1.13727572858934e-05,
|
| 584 |
+
"loss": 0.2766,
|
| 585 |
+
"num_tokens": 436166148.0,
|
| 586 |
+
"step": 72
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 3.6666666666666665,
|
| 590 |
+
"grad_norm": 0.07330776888938495,
|
| 591 |
+
"learning_rate": 1.0907887619183308e-05,
|
| 592 |
+
"loss": 0.2815,
|
| 593 |
+
"num_tokens": 442404416.0,
|
| 594 |
+
"step": 73
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 3.717948717948718,
|
| 598 |
+
"grad_norm": 0.07750844403333255,
|
| 599 |
+
"learning_rate": 1.0454652046639486e-05,
|
| 600 |
+
"loss": 0.2765,
|
| 601 |
+
"num_tokens": 448671262.0,
|
| 602 |
+
"step": 74
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 3.769230769230769,
|
| 606 |
+
"grad_norm": 0.06910018245510882,
|
| 607 |
+
"learning_rate": 1.0013525949789745e-05,
|
| 608 |
+
"loss": 0.2765,
|
| 609 |
+
"num_tokens": 454908778.0,
|
| 610 |
+
"step": 75
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 3.8205128205128203,
|
| 614 |
+
"grad_norm": 0.06772004012970233,
|
| 615 |
+
"learning_rate": 9.584972008994123e-06,
|
| 616 |
+
"loss": 0.2761,
|
| 617 |
+
"num_tokens": 461176026.0,
|
| 618 |
+
"step": 76
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 3.871794871794872,
|
| 622 |
+
"grad_norm": 0.06449052728723467,
|
| 623 |
+
"learning_rate": 9.16943971815708e-06,
|
| 624 |
+
"loss": 0.2718,
|
| 625 |
+
"num_tokens": 467459020.0,
|
| 626 |
+
"step": 77
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 3.9230769230769234,
|
| 630 |
+
"grad_norm": 0.06591642004614524,
|
| 631 |
+
"learning_rate": 8.767364913270399e-06,
|
| 632 |
+
"loss": 0.2686,
|
| 633 |
+
"num_tokens": 473721944.0,
|
| 634 |
+
"step": 78
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 3.9743589743589745,
|
| 638 |
+
"grad_norm": 0.05776576407058006,
|
| 639 |
+
"learning_rate": 8.379169315281485e-06,
|
| 640 |
+
"loss": 0.2763,
|
| 641 |
+
"num_tokens": 480002563.0,
|
| 642 |
+
"step": 79
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 4.0,
|
| 646 |
+
"grad_norm": 0.13745363566927554,
|
| 647 |
+
"learning_rate": 8.005260087766318e-06,
|
| 648 |
+
"loss": 0.268,
|
| 649 |
+
"num_tokens": 481441850.0,
|
| 650 |
+
"step": 80
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 4.051282051282051,
|
| 654 |
+
"grad_norm": 0.09534028423404649,
|
| 655 |
+
"learning_rate": 7.646029409871029e-06,
|
| 656 |
+
"loss": 0.2577,
|
| 657 |
+
"num_tokens": 487660172.0,
|
| 658 |
+
"step": 81
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 4.102564102564102,
|
| 662 |
+
"grad_norm": 0.06997274489921491,
|
| 663 |
+
"learning_rate": 7.301854064970202e-06,
|
| 664 |
+
"loss": 0.2557,
|
| 665 |
+
"num_tokens": 493912073.0,
|
| 666 |
+
"step": 82
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 4.153846153846154,
|
| 670 |
+
"grad_norm": 0.08755702656310224,
|
| 671 |
+
"learning_rate": 6.973095045473124e-06,
|
| 672 |
+
"loss": 0.2604,
|
| 673 |
+
"num_tokens": 500168012.0,
|
| 674 |
+
"step": 83
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 4.205128205128205,
|
| 678 |
+
"grad_norm": 0.08888143272022463,
|
| 679 |
+
"learning_rate": 6.660097174192556e-06,
|
| 680 |
+
"loss": 0.2629,
|
| 681 |
+
"num_tokens": 506448923.0,
|
| 682 |
+
"step": 84
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 4.256410256410256,
|
| 686 |
+
"grad_norm": 0.06765618005674934,
|
| 687 |
+
"learning_rate": 6.363188742673281e-06,
|
| 688 |
+
"loss": 0.2597,
|
| 689 |
+
"num_tokens": 512697543.0,
|
| 690 |
+
"step": 85
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 4.3076923076923075,
|
| 694 |
+
"grad_norm": 0.08077914878872741,
|
| 695 |
+
"learning_rate": 6.082681166859579e-06,
|
| 696 |
+
"loss": 0.2565,
|
| 697 |
+
"num_tokens": 518952837.0,
|
| 698 |
+
"step": 86
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 4.358974358974359,
|
| 702 |
+
"grad_norm": 0.07580797752338386,
|
| 703 |
+
"learning_rate": 5.818868660462886e-06,
|
| 704 |
+
"loss": 0.2622,
|
| 705 |
+
"num_tokens": 525234697.0,
|
| 706 |
+
"step": 87
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 4.410256410256411,
|
| 710 |
+
"grad_norm": 0.06618677478470034,
|
| 711 |
+
"learning_rate": 5.5720279263722795e-06,
|
| 712 |
+
"loss": 0.2571,
|
| 713 |
+
"num_tokens": 531495708.0,
|
| 714 |
+
"step": 88
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 4.461538461538462,
|
| 718 |
+
"grad_norm": 0.0680578770532582,
|
| 719 |
+
"learning_rate": 5.342417866431326e-06,
|
| 720 |
+
"loss": 0.2583,
|
| 721 |
+
"num_tokens": 537744492.0,
|
| 722 |
+
"step": 89
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 4.512820512820513,
|
| 726 |
+
"grad_norm": 0.0730363106871515,
|
| 727 |
+
"learning_rate": 5.130279309885817e-06,
|
| 728 |
+
"loss": 0.2523,
|
| 729 |
+
"num_tokens": 544012162.0,
|
| 730 |
+
"step": 90
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 4.564102564102564,
|
| 734 |
+
"grad_norm": 0.06548896520491608,
|
| 735 |
+
"learning_rate": 4.935834760787133e-06,
|
| 736 |
+
"loss": 0.2556,
|
| 737 |
+
"num_tokens": 550280847.0,
|
| 738 |
+
"step": 91
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 4.615384615384615,
|
| 742 |
+
"grad_norm": 0.06057215331898529,
|
| 743 |
+
"learning_rate": 4.7592881646162336e-06,
|
| 744 |
+
"loss": 0.2534,
|
| 745 |
+
"num_tokens": 556540016.0,
|
| 746 |
+
"step": 92
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 4.666666666666667,
|
| 750 |
+
"grad_norm": 0.0660973563878409,
|
| 751 |
+
"learning_rate": 4.600824694373e-06,
|
| 752 |
+
"loss": 0.259,
|
| 753 |
+
"num_tokens": 562819807.0,
|
| 754 |
+
"step": 93
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 4.717948717948718,
|
| 758 |
+
"grad_norm": 0.06127804475677186,
|
| 759 |
+
"learning_rate": 4.460610556355333e-06,
|
| 760 |
+
"loss": 0.2516,
|
| 761 |
+
"num_tokens": 569101952.0,
|
| 762 |
+
"step": 94
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 4.769230769230769,
|
| 766 |
+
"grad_norm": 0.06055474487432341,
|
| 767 |
+
"learning_rate": 4.338792815831698e-06,
|
| 768 |
+
"loss": 0.2496,
|
| 769 |
+
"num_tokens": 575336966.0,
|
| 770 |
+
"step": 95
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 4.82051282051282,
|
| 774 |
+
"grad_norm": 0.05900101598712522,
|
| 775 |
+
"learning_rate": 4.2354992427899674e-06,
|
| 776 |
+
"loss": 0.2525,
|
| 777 |
+
"num_tokens": 581620237.0,
|
| 778 |
+
"step": 96
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 4.871794871794872,
|
| 782 |
+
"grad_norm": 0.06942816751525951,
|
| 783 |
+
"learning_rate": 4.150838177924349e-06,
|
| 784 |
+
"loss": 0.2589,
|
| 785 |
+
"num_tokens": 587875142.0,
|
| 786 |
+
"step": 97
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 4.923076923076923,
|
| 790 |
+
"grad_norm": 0.06137215301815073,
|
| 791 |
+
"learning_rate": 4.0848984190009495e-06,
|
| 792 |
+
"loss": 0.2616,
|
| 793 |
+
"num_tokens": 594142883.0,
|
| 794 |
+
"step": 98
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 4.9743589743589745,
|
| 798 |
+
"grad_norm": 0.06146449383500362,
|
| 799 |
+
"learning_rate": 4.037749127721191e-06,
|
| 800 |
+
"loss": 0.2508,
|
| 801 |
+
"num_tokens": 600408547.0,
|
| 802 |
+
"step": 99
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 5.0,
|
| 806 |
+
"grad_norm": 0.06146449383500362,
|
| 807 |
+
"learning_rate": 4.009439757180732e-06,
|
| 808 |
+
"loss": 0.247,
|
| 809 |
+
"num_tokens": 601802442.0,
|
| 810 |
+
"step": 100
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 5.0,
|
| 814 |
+
"step": 100,
|
| 815 |
+
"total_flos": 2.496956368699785e+18,
|
| 816 |
+
"train_loss": 0.33275703743100166,
|
| 817 |
+
"train_runtime": 21613.1094,
|
| 818 |
+
"train_samples_per_second": 0.863,
|
| 819 |
+
"train_steps_per_second": 0.005
|
| 820 |
}
|
| 821 |
],
|
| 822 |
"logging_steps": 1,
|
| 823 |
+
"max_steps": 100,
|
| 824 |
"num_input_tokens_seen": 0,
|
| 825 |
+
"num_train_epochs": 5,
|
| 826 |
+
"save_steps": 5,
|
| 827 |
"stateful_callbacks": {
|
| 828 |
"TrainerControl": {
|
| 829 |
"args": {
|
|
|
|
| 836 |
"attributes": {}
|
| 837 |
}
|
| 838 |
},
|
| 839 |
+
"total_flos": 2.496956368699785e+18,
|
| 840 |
"train_batch_size": 1,
|
| 841 |
"trial_name": null,
|
| 842 |
"trial_params": null
|