End of training
Browse files- all_results.json +16 -0
- eval_results.json +11 -0
- runs/Mar17_11-31-03_d7c13a14123e/events.out.tfevents.1742215636.d7c13a14123e.3775.7 +3 -0
- train_results.json +8 -0
- trainer_state.json +796 -0
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 17.905426356589146,
|
| 3 |
+
"eval_accuracy": 0.9597467209407508,
|
| 4 |
+
"eval_f1": 0.9595276897831222,
|
| 5 |
+
"eval_loss": 0.11270730942487717,
|
| 6 |
+
"eval_precision": 0.9601617456499858,
|
| 7 |
+
"eval_recall": 0.9597467209407508,
|
| 8 |
+
"eval_runtime": 30.7679,
|
| 9 |
+
"eval_samples_per_second": 71.861,
|
| 10 |
+
"eval_steps_per_second": 9.003,
|
| 11 |
+
"total_flos": 1.0232905043346113e+19,
|
| 12 |
+
"train_loss": 0.2441097328580659,
|
| 13 |
+
"train_runtime": 4508.9942,
|
| 14 |
+
"train_samples_per_second": 41.182,
|
| 15 |
+
"train_steps_per_second": 1.285
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 17.905426356589146,
|
| 3 |
+
"eval_accuracy": 0.9597467209407508,
|
| 4 |
+
"eval_f1": 0.9595276897831222,
|
| 5 |
+
"eval_loss": 0.11270730942487717,
|
| 6 |
+
"eval_precision": 0.9601617456499858,
|
| 7 |
+
"eval_recall": 0.9597467209407508,
|
| 8 |
+
"eval_runtime": 30.7679,
|
| 9 |
+
"eval_samples_per_second": 71.861,
|
| 10 |
+
"eval_steps_per_second": 9.003
|
| 11 |
+
}
|
runs/Mar17_11-31-03_d7c13a14123e/events.out.tfevents.1742215636.d7c13a14123e.3775.7
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1a8d5cc3bdf09e4074704ab09cdc9ab1b78265f3fb106c293d739b3dbcbe1c39
|
| 3 |
+
size 1032
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 17.905426356589146,
|
| 3 |
+
"total_flos": 1.0232905043346113e+19,
|
| 4 |
+
"train_loss": 0.2441097328580659,
|
| 5 |
+
"train_runtime": 4508.9942,
|
| 6 |
+
"train_samples_per_second": 41.182,
|
| 7 |
+
"train_steps_per_second": 1.285
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,796 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9595276897831222,
|
| 3 |
+
"best_model_checkpoint": "dino-vitb8-finetuned-stroke-binary/checkpoint-2900",
|
| 4 |
+
"epoch": 17.905426356589146,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 2900,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.31007751937984496,
|
| 13 |
+
"grad_norm": 6.24950647354126,
|
| 14 |
+
"learning_rate": 1.6896551724137933e-06,
|
| 15 |
+
"loss": 0.8366,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.6201550387596899,
|
| 20 |
+
"grad_norm": 5.115139007568359,
|
| 21 |
+
"learning_rate": 3.4137931034482764e-06,
|
| 22 |
+
"loss": 0.7965,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.6201550387596899,
|
| 27 |
+
"eval_accuracy": 0.5382180009045681,
|
| 28 |
+
"eval_f1": 0.5057986540639501,
|
| 29 |
+
"eval_loss": 0.8312332630157471,
|
| 30 |
+
"eval_precision": 0.496820169669259,
|
| 31 |
+
"eval_recall": 0.5382180009045681,
|
| 32 |
+
"eval_runtime": 32.0161,
|
| 33 |
+
"eval_samples_per_second": 69.059,
|
| 34 |
+
"eval_steps_per_second": 8.652,
|
| 35 |
+
"step": 100
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 0.9302325581395349,
|
| 39 |
+
"grad_norm": 8.835868835449219,
|
| 40 |
+
"learning_rate": 5.137931034482759e-06,
|
| 41 |
+
"loss": 0.7713,
|
| 42 |
+
"step": 150
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 1.235658914728682,
|
| 46 |
+
"grad_norm": 9.565229415893555,
|
| 47 |
+
"learning_rate": 6.862068965517242e-06,
|
| 48 |
+
"loss": 0.6839,
|
| 49 |
+
"step": 200
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 1.235658914728682,
|
| 53 |
+
"eval_accuracy": 0.6246042514699232,
|
| 54 |
+
"eval_f1": 0.5749818840100107,
|
| 55 |
+
"eval_loss": 0.6796302199363708,
|
| 56 |
+
"eval_precision": 0.5990946901610464,
|
| 57 |
+
"eval_recall": 0.6246042514699232,
|
| 58 |
+
"eval_runtime": 32.1198,
|
| 59 |
+
"eval_samples_per_second": 68.836,
|
| 60 |
+
"eval_steps_per_second": 8.624,
|
| 61 |
+
"step": 200
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 1.5457364341085271,
|
| 65 |
+
"grad_norm": 8.520139694213867,
|
| 66 |
+
"learning_rate": 8.586206896551726e-06,
|
| 67 |
+
"loss": 0.6219,
|
| 68 |
+
"step": 250
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 1.8558139534883722,
|
| 72 |
+
"grad_norm": 6.877398490905762,
|
| 73 |
+
"learning_rate": 1.0310344827586208e-05,
|
| 74 |
+
"loss": 0.5662,
|
| 75 |
+
"step": 300
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 1.8558139534883722,
|
| 79 |
+
"eval_accuracy": 0.7317955676164631,
|
| 80 |
+
"eval_f1": 0.7119348319857804,
|
| 81 |
+
"eval_loss": 0.5344149470329285,
|
| 82 |
+
"eval_precision": 0.7376504962044594,
|
| 83 |
+
"eval_recall": 0.7317955676164631,
|
| 84 |
+
"eval_runtime": 32.0121,
|
| 85 |
+
"eval_samples_per_second": 69.068,
|
| 86 |
+
"eval_steps_per_second": 8.653,
|
| 87 |
+
"step": 300
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 2.1612403100775195,
|
| 91 |
+
"grad_norm": 6.749499797821045,
|
| 92 |
+
"learning_rate": 1.203448275862069e-05,
|
| 93 |
+
"loss": 0.4908,
|
| 94 |
+
"step": 350
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 2.471317829457364,
|
| 98 |
+
"grad_norm": 11.308539390563965,
|
| 99 |
+
"learning_rate": 1.3758620689655173e-05,
|
| 100 |
+
"loss": 0.4408,
|
| 101 |
+
"step": 400
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 2.471317829457364,
|
| 105 |
+
"eval_accuracy": 0.8123021257349615,
|
| 106 |
+
"eval_f1": 0.8082421593111531,
|
| 107 |
+
"eval_loss": 0.40815532207489014,
|
| 108 |
+
"eval_precision": 0.8120443114553312,
|
| 109 |
+
"eval_recall": 0.8123021257349615,
|
| 110 |
+
"eval_runtime": 31.919,
|
| 111 |
+
"eval_samples_per_second": 69.269,
|
| 112 |
+
"eval_steps_per_second": 8.678,
|
| 113 |
+
"step": 400
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 2.781395348837209,
|
| 117 |
+
"grad_norm": 3.188601493835449,
|
| 118 |
+
"learning_rate": 1.548275862068966e-05,
|
| 119 |
+
"loss": 0.3761,
|
| 120 |
+
"step": 450
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 3.0868217054263565,
|
| 124 |
+
"grad_norm": 9.05437183380127,
|
| 125 |
+
"learning_rate": 1.7206896551724138e-05,
|
| 126 |
+
"loss": 0.3611,
|
| 127 |
+
"step": 500
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"epoch": 3.0868217054263565,
|
| 131 |
+
"eval_accuracy": 0.8602442333785617,
|
| 132 |
+
"eval_f1": 0.859700206427454,
|
| 133 |
+
"eval_loss": 0.333509624004364,
|
| 134 |
+
"eval_precision": 0.8595584495556567,
|
| 135 |
+
"eval_recall": 0.8602442333785617,
|
| 136 |
+
"eval_runtime": 32.0233,
|
| 137 |
+
"eval_samples_per_second": 69.044,
|
| 138 |
+
"eval_steps_per_second": 8.65,
|
| 139 |
+
"step": 500
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 3.3968992248062015,
|
| 143 |
+
"grad_norm": 2.100419759750366,
|
| 144 |
+
"learning_rate": 1.893103448275862e-05,
|
| 145 |
+
"loss": 0.3072,
|
| 146 |
+
"step": 550
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 3.7069767441860466,
|
| 150 |
+
"grad_norm": 6.307981014251709,
|
| 151 |
+
"learning_rate": 1.9999345217525786e-05,
|
| 152 |
+
"loss": 0.3121,
|
| 153 |
+
"step": 600
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 3.7069767441860466,
|
| 157 |
+
"eval_accuracy": 0.8860244233378561,
|
| 158 |
+
"eval_f1": 0.8831883391472106,
|
| 159 |
+
"eval_loss": 0.27464836835861206,
|
| 160 |
+
"eval_precision": 0.8913973600585926,
|
| 161 |
+
"eval_recall": 0.8860244233378561,
|
| 162 |
+
"eval_runtime": 31.9786,
|
| 163 |
+
"eval_samples_per_second": 69.14,
|
| 164 |
+
"eval_steps_per_second": 8.662,
|
| 165 |
+
"step": 600
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 4.0124031007751935,
|
| 169 |
+
"grad_norm": 4.161785125732422,
|
| 170 |
+
"learning_rate": 1.999136563788985e-05,
|
| 171 |
+
"loss": 0.2786,
|
| 172 |
+
"step": 650
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 4.322480620155039,
|
| 176 |
+
"grad_norm": 5.515849590301514,
|
| 177 |
+
"learning_rate": 1.997432546533359e-05,
|
| 178 |
+
"loss": 0.2614,
|
| 179 |
+
"step": 700
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 4.322480620155039,
|
| 183 |
+
"eval_accuracy": 0.9050203527815468,
|
| 184 |
+
"eval_f1": 0.9039775360154476,
|
| 185 |
+
"eval_loss": 0.2299066036939621,
|
| 186 |
+
"eval_precision": 0.90576736213102,
|
| 187 |
+
"eval_recall": 0.9050203527815468,
|
| 188 |
+
"eval_runtime": 31.9694,
|
| 189 |
+
"eval_samples_per_second": 69.16,
|
| 190 |
+
"eval_steps_per_second": 8.665,
|
| 191 |
+
"step": 700
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 4.632558139534884,
|
| 195 |
+
"grad_norm": 2.6630172729492188,
|
| 196 |
+
"learning_rate": 1.9948240152606154e-05,
|
| 197 |
+
"loss": 0.2622,
|
| 198 |
+
"step": 750
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 4.942635658914728,
|
| 202 |
+
"grad_norm": 4.852052211761475,
|
| 203 |
+
"learning_rate": 1.9913133354972336e-05,
|
| 204 |
+
"loss": 0.242,
|
| 205 |
+
"step": 800
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 4.942635658914728,
|
| 209 |
+
"eval_accuracy": 0.9176843057440073,
|
| 210 |
+
"eval_f1": 0.9177728397549259,
|
| 211 |
+
"eval_loss": 0.21027454733848572,
|
| 212 |
+
"eval_precision": 0.9179004728816644,
|
| 213 |
+
"eval_recall": 0.9176843057440073,
|
| 214 |
+
"eval_runtime": 31.9231,
|
| 215 |
+
"eval_samples_per_second": 69.26,
|
| 216 |
+
"eval_steps_per_second": 8.677,
|
| 217 |
+
"step": 800
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 5.248062015503876,
|
| 221 |
+
"grad_norm": 5.338088035583496,
|
| 222 |
+
"learning_rate": 1.9869036908760982e-05,
|
| 223 |
+
"loss": 0.2332,
|
| 224 |
+
"step": 850
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 5.558139534883721,
|
| 228 |
+
"grad_norm": 1.9332038164138794,
|
| 229 |
+
"learning_rate": 1.981599080249447e-05,
|
| 230 |
+
"loss": 0.2239,
|
| 231 |
+
"step": 900
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 5.558139534883721,
|
| 235 |
+
"eval_accuracy": 0.9081863410221619,
|
| 236 |
+
"eval_f1": 0.908987571769516,
|
| 237 |
+
"eval_loss": 0.22980067133903503,
|
| 238 |
+
"eval_precision": 0.9136294473473457,
|
| 239 |
+
"eval_recall": 0.9081863410221619,
|
| 240 |
+
"eval_runtime": 31.9612,
|
| 241 |
+
"eval_samples_per_second": 69.178,
|
| 242 |
+
"eval_steps_per_second": 8.667,
|
| 243 |
+
"step": 900
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 5.868217054263566,
|
| 247 |
+
"grad_norm": 4.6057281494140625,
|
| 248 |
+
"learning_rate": 1.9754043140625415e-05,
|
| 249 |
+
"loss": 0.2216,
|
| 250 |
+
"step": 950
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 6.173643410852713,
|
| 254 |
+
"grad_norm": 3.815474271774292,
|
| 255 |
+
"learning_rate": 1.9683250099913514e-05,
|
| 256 |
+
"loss": 0.1979,
|
| 257 |
+
"step": 1000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 6.173643410852713,
|
| 261 |
+
"eval_accuracy": 0.9208502939846224,
|
| 262 |
+
"eval_f1": 0.9197449979522355,
|
| 263 |
+
"eval_loss": 0.20585399866104126,
|
| 264 |
+
"eval_precision": 0.9230342997728979,
|
| 265 |
+
"eval_recall": 0.9208502939846224,
|
| 266 |
+
"eval_runtime": 32.0879,
|
| 267 |
+
"eval_samples_per_second": 68.904,
|
| 268 |
+
"eval_steps_per_second": 8.633,
|
| 269 |
+
"step": 1000
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 6.4837209302325585,
|
| 273 |
+
"grad_norm": 3.7233753204345703,
|
| 274 |
+
"learning_rate": 1.960367587848211e-05,
|
| 275 |
+
"loss": 0.2074,
|
| 276 |
+
"step": 1050
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 6.793798449612403,
|
| 280 |
+
"grad_norm": 3.686628580093384,
|
| 281 |
+
"learning_rate": 1.9515392637600583e-05,
|
| 282 |
+
"loss": 0.2082,
|
| 283 |
+
"step": 1100
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 6.793798449612403,
|
| 287 |
+
"eval_accuracy": 0.9262777023971054,
|
| 288 |
+
"eval_f1": 0.926117263289722,
|
| 289 |
+
"eval_loss": 0.17793285846710205,
|
| 290 |
+
"eval_precision": 0.9261011200078554,
|
| 291 |
+
"eval_recall": 0.9262777023971054,
|
| 292 |
+
"eval_runtime": 31.9195,
|
| 293 |
+
"eval_samples_per_second": 69.268,
|
| 294 |
+
"eval_steps_per_second": 8.678,
|
| 295 |
+
"step": 1100
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 7.09922480620155,
|
| 299 |
+
"grad_norm": 3.4548275470733643,
|
| 300 |
+
"learning_rate": 1.9418480436245487e-05,
|
| 301 |
+
"loss": 0.1998,
|
| 302 |
+
"step": 1150
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 7.409302325581395,
|
| 306 |
+
"grad_norm": 1.4032052755355835,
|
| 307 |
+
"learning_rate": 1.931302715849967e-05,
|
| 308 |
+
"loss": 0.1723,
|
| 309 |
+
"step": 1200
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 7.409302325581395,
|
| 313 |
+
"eval_accuracy": 0.9308005427408412,
|
| 314 |
+
"eval_f1": 0.9301945813993165,
|
| 315 |
+
"eval_loss": 0.1693263202905655,
|
| 316 |
+
"eval_precision": 0.9315485344818776,
|
| 317 |
+
"eval_recall": 0.9308005427408412,
|
| 318 |
+
"eval_runtime": 31.8606,
|
| 319 |
+
"eval_samples_per_second": 69.396,
|
| 320 |
+
"eval_steps_per_second": 8.694,
|
| 321 |
+
"step": 1200
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 7.7193798449612405,
|
| 325 |
+
"grad_norm": 2.225787401199341,
|
| 326 |
+
"learning_rate": 1.9199128433855277e-05,
|
| 327 |
+
"loss": 0.1864,
|
| 328 |
+
"step": 1250
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 8.024806201550387,
|
| 332 |
+
"grad_norm": 4.083457946777344,
|
| 333 |
+
"learning_rate": 1.9076887550492877e-05,
|
| 334 |
+
"loss": 0.1877,
|
| 335 |
+
"step": 1300
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 8.024806201550387,
|
| 339 |
+
"eval_accuracy": 0.9380370872908186,
|
| 340 |
+
"eval_f1": 0.9381736344846147,
|
| 341 |
+
"eval_loss": 0.168083056807518,
|
| 342 |
+
"eval_precision": 0.9384861550793825,
|
| 343 |
+
"eval_recall": 0.9380370872908186,
|
| 344 |
+
"eval_runtime": 32.0939,
|
| 345 |
+
"eval_samples_per_second": 68.892,
|
| 346 |
+
"eval_steps_per_second": 8.631,
|
| 347 |
+
"step": 1300
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 8.334883720930232,
|
| 351 |
+
"grad_norm": 5.380185127258301,
|
| 352 |
+
"learning_rate": 1.894641536161537e-05,
|
| 353 |
+
"loss": 0.1662,
|
| 354 |
+
"step": 1350
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 8.644961240310078,
|
| 358 |
+
"grad_norm": 2.990643262863159,
|
| 359 |
+
"learning_rate": 1.8807830184921636e-05,
|
| 360 |
+
"loss": 0.2,
|
| 361 |
+
"step": 1400
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 8.644961240310078,
|
| 365 |
+
"eval_accuracy": 0.9402985074626866,
|
| 366 |
+
"eval_f1": 0.9401756807643212,
|
| 367 |
+
"eval_loss": 0.14824680984020233,
|
| 368 |
+
"eval_precision": 0.9401845264674767,
|
| 369 |
+
"eval_recall": 0.9402985074626866,
|
| 370 |
+
"eval_runtime": 31.8892,
|
| 371 |
+
"eval_samples_per_second": 69.334,
|
| 372 |
+
"eval_steps_per_second": 8.686,
|
| 373 |
+
"step": 1400
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 8.955038759689922,
|
| 377 |
+
"grad_norm": 3.6064834594726562,
|
| 378 |
+
"learning_rate": 1.8661257695311006e-05,
|
| 379 |
+
"loss": 0.1621,
|
| 380 |
+
"step": 1450
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 9.26046511627907,
|
| 384 |
+
"grad_norm": 5.486879825592041,
|
| 385 |
+
"learning_rate": 1.8506830810915983e-05,
|
| 386 |
+
"loss": 0.1642,
|
| 387 |
+
"step": 1500
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 9.26046511627907,
|
| 391 |
+
"eval_accuracy": 0.9330619629127092,
|
| 392 |
+
"eval_f1": 0.9322023134638141,
|
| 393 |
+
"eval_loss": 0.16374026238918304,
|
| 394 |
+
"eval_precision": 0.9352048827585053,
|
| 395 |
+
"eval_recall": 0.9330619629127092,
|
| 396 |
+
"eval_runtime": 32.0395,
|
| 397 |
+
"eval_samples_per_second": 69.009,
|
| 398 |
+
"eval_steps_per_second": 8.646,
|
| 399 |
+
"step": 1500
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 9.570542635658915,
|
| 403 |
+
"grad_norm": 2.025575637817383,
|
| 404 |
+
"learning_rate": 1.8344689572566436e-05,
|
| 405 |
+
"loss": 0.1586,
|
| 406 |
+
"step": 1550
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 9.88062015503876,
|
| 410 |
+
"grad_norm": 3.0411391258239746,
|
| 411 |
+
"learning_rate": 1.8174981016794653e-05,
|
| 412 |
+
"loss": 0.1525,
|
| 413 |
+
"step": 1600
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 9.88062015503876,
|
| 417 |
+
"eval_accuracy": 0.942107643600181,
|
| 418 |
+
"eval_f1": 0.94173840230828,
|
| 419 |
+
"eval_loss": 0.14938268065452576,
|
| 420 |
+
"eval_precision": 0.9424881755892641,
|
| 421 |
+
"eval_recall": 0.942107643600181,
|
| 422 |
+
"eval_runtime": 31.8834,
|
| 423 |
+
"eval_samples_per_second": 69.347,
|
| 424 |
+
"eval_steps_per_second": 8.688,
|
| 425 |
+
"step": 1600
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 10.186046511627907,
|
| 429 |
+
"grad_norm": 3.4836244583129883,
|
| 430 |
+
"learning_rate": 1.799785904249642e-05,
|
| 431 |
+
"loss": 0.1649,
|
| 432 |
+
"step": 1650
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 10.496124031007753,
|
| 436 |
+
"grad_norm": 3.605983257293701,
|
| 437 |
+
"learning_rate": 1.7813484271368947e-05,
|
| 438 |
+
"loss": 0.158,
|
| 439 |
+
"step": 1700
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 10.496124031007753,
|
| 443 |
+
"eval_accuracy": 0.9484396200814111,
|
| 444 |
+
"eval_f1": 0.947995545399902,
|
| 445 |
+
"eval_loss": 0.1402619332075119,
|
| 446 |
+
"eval_precision": 0.9494961047542965,
|
| 447 |
+
"eval_recall": 0.9484396200814111,
|
| 448 |
+
"eval_runtime": 32.0182,
|
| 449 |
+
"eval_samples_per_second": 69.055,
|
| 450 |
+
"eval_steps_per_second": 8.651,
|
| 451 |
+
"step": 1700
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 10.806201550387597,
|
| 455 |
+
"grad_norm": 4.386783123016357,
|
| 456 |
+
"learning_rate": 1.7622023902252378e-05,
|
| 457 |
+
"loss": 0.1669,
|
| 458 |
+
"step": 1750
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 11.111627906976745,
|
| 462 |
+
"grad_norm": 3.84214186668396,
|
| 463 |
+
"learning_rate": 1.742365155950676e-05,
|
| 464 |
+
"loss": 0.1327,
|
| 465 |
+
"step": 1800
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 11.111627906976745,
|
| 469 |
+
"eval_accuracy": 0.9497964721845319,
|
| 470 |
+
"eval_f1": 0.9498129231725488,
|
| 471 |
+
"eval_loss": 0.13292284309864044,
|
| 472 |
+
"eval_precision": 0.9498330407980847,
|
| 473 |
+
"eval_recall": 0.9497964721845319,
|
| 474 |
+
"eval_runtime": 32.0593,
|
| 475 |
+
"eval_samples_per_second": 68.966,
|
| 476 |
+
"eval_steps_per_second": 8.64,
|
| 477 |
+
"step": 1800
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 11.421705426356588,
|
| 481 |
+
"grad_norm": 2.3866167068481445,
|
| 482 |
+
"learning_rate": 1.7218547135562155e-05,
|
| 483 |
+
"loss": 0.1679,
|
| 484 |
+
"step": 1850
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 11.731782945736434,
|
| 488 |
+
"grad_norm": 5.185602188110352,
|
| 489 |
+
"learning_rate": 1.700689662778458e-05,
|
| 490 |
+
"loss": 0.1465,
|
| 491 |
+
"step": 1900
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 11.731782945736434,
|
| 495 |
+
"eval_accuracy": 0.9525101763907734,
|
| 496 |
+
"eval_f1": 0.9523724534109135,
|
| 497 |
+
"eval_loss": 0.12334821373224258,
|
| 498 |
+
"eval_precision": 0.9524932751901988,
|
| 499 |
+
"eval_recall": 0.9525101763907734,
|
| 500 |
+
"eval_runtime": 31.92,
|
| 501 |
+
"eval_samples_per_second": 69.267,
|
| 502 |
+
"eval_steps_per_second": 8.678,
|
| 503 |
+
"step": 1900
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 12.037209302325582,
|
| 507 |
+
"grad_norm": 9.162732124328613,
|
| 508 |
+
"learning_rate": 1.678889196980572e-05,
|
| 509 |
+
"loss": 0.1291,
|
| 510 |
+
"step": 1950
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 12.347286821705426,
|
| 514 |
+
"grad_norm": 5.636198043823242,
|
| 515 |
+
"learning_rate": 1.6564730857469392e-05,
|
| 516 |
+
"loss": 0.1311,
|
| 517 |
+
"step": 2000
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 12.347286821705426,
|
| 521 |
+
"eval_accuracy": 0.9520578923563998,
|
| 522 |
+
"eval_f1": 0.9520149841974469,
|
| 523 |
+
"eval_loss": 0.12798894941806793,
|
| 524 |
+
"eval_precision": 0.9519983389426212,
|
| 525 |
+
"eval_recall": 0.9520578923563998,
|
| 526 |
+
"eval_runtime": 31.9866,
|
| 527 |
+
"eval_samples_per_second": 69.123,
|
| 528 |
+
"eval_steps_per_second": 8.66,
|
| 529 |
+
"step": 2000
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 12.657364341085271,
|
| 533 |
+
"grad_norm": 2.2612485885620117,
|
| 534 |
+
"learning_rate": 1.633461656955259e-05,
|
| 535 |
+
"loss": 0.1619,
|
| 536 |
+
"step": 2050
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 12.967441860465117,
|
| 540 |
+
"grad_norm": 2.522465229034424,
|
| 541 |
+
"learning_rate": 1.609875778342372e-05,
|
| 542 |
+
"loss": 0.129,
|
| 543 |
+
"step": 2100
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 12.967441860465117,
|
| 547 |
+
"eval_accuracy": 0.9556761646313885,
|
| 548 |
+
"eval_f1": 0.9555744118131845,
|
| 549 |
+
"eval_loss": 0.11725445836782455,
|
| 550 |
+
"eval_precision": 0.9556383737245242,
|
| 551 |
+
"eval_recall": 0.9556761646313885,
|
| 552 |
+
"eval_runtime": 31.8471,
|
| 553 |
+
"eval_samples_per_second": 69.426,
|
| 554 |
+
"eval_steps_per_second": 8.698,
|
| 555 |
+
"step": 2100
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 13.272868217054263,
|
| 559 |
+
"grad_norm": 7.201235294342041,
|
| 560 |
+
"learning_rate": 1.5857368385805087e-05,
|
| 561 |
+
"loss": 0.1351,
|
| 562 |
+
"step": 2150
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 13.582945736434109,
|
| 566 |
+
"grad_norm": 4.276857852935791,
|
| 567 |
+
"learning_rate": 1.561565191643536e-05,
|
| 568 |
+
"loss": 0.1425,
|
| 569 |
+
"step": 2200
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 13.582945736434109,
|
| 573 |
+
"eval_accuracy": 0.9552238805970149,
|
| 574 |
+
"eval_f1": 0.9552090145107874,
|
| 575 |
+
"eval_loss": 0.11896734684705734,
|
| 576 |
+
"eval_precision": 0.9551978476090869,
|
| 577 |
+
"eval_recall": 0.9552238805970149,
|
| 578 |
+
"eval_runtime": 31.9931,
|
| 579 |
+
"eval_samples_per_second": 69.109,
|
| 580 |
+
"eval_steps_per_second": 8.658,
|
| 581 |
+
"step": 2200
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 13.893023255813954,
|
| 585 |
+
"grad_norm": 3.126692295074463,
|
| 586 |
+
"learning_rate": 1.5363962356243304e-05,
|
| 587 |
+
"loss": 0.1249,
|
| 588 |
+
"step": 2250
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 14.1984496124031,
|
| 592 |
+
"grad_norm": 2.0163676738739014,
|
| 593 |
+
"learning_rate": 1.5107408528131196e-05,
|
| 594 |
+
"loss": 0.1256,
|
| 595 |
+
"step": 2300
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 14.1984496124031,
|
| 599 |
+
"eval_accuracy": 0.9565807327001357,
|
| 600 |
+
"eval_f1": 0.9563271746916711,
|
| 601 |
+
"eval_loss": 0.12249390035867691,
|
| 602 |
+
"eval_precision": 0.9570420286206239,
|
| 603 |
+
"eval_recall": 0.9565807327001357,
|
| 604 |
+
"eval_runtime": 32.0483,
|
| 605 |
+
"eval_samples_per_second": 68.99,
|
| 606 |
+
"eval_steps_per_second": 8.643,
|
| 607 |
+
"step": 2300
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 14.508527131782946,
|
| 611 |
+
"grad_norm": 2.3576133251190186,
|
| 612 |
+
"learning_rate": 1.4846223085960681e-05,
|
| 613 |
+
"loss": 0.1331,
|
| 614 |
+
"step": 2350
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 14.81860465116279,
|
| 618 |
+
"grad_norm": 3.005863666534424,
|
| 619 |
+
"learning_rate": 1.45806428837369e-05,
|
| 620 |
+
"loss": 0.1461,
|
| 621 |
+
"step": 2400
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 14.81860465116279,
|
| 625 |
+
"eval_accuracy": 0.9588421528720036,
|
| 626 |
+
"eval_f1": 0.9587720391887182,
|
| 627 |
+
"eval_loss": 0.11707326769828796,
|
| 628 |
+
"eval_precision": 0.9587952797993504,
|
| 629 |
+
"eval_recall": 0.9588421528720036,
|
| 630 |
+
"eval_runtime": 31.9123,
|
| 631 |
+
"eval_samples_per_second": 69.284,
|
| 632 |
+
"eval_steps_per_second": 8.68,
|
| 633 |
+
"step": 2400
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 15.124031007751938,
|
| 637 |
+
"grad_norm": 2.8450655937194824,
|
| 638 |
+
"learning_rate": 1.4310908760819285e-05,
|
| 639 |
+
"loss": 0.1193,
|
| 640 |
+
"step": 2450
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 15.434108527131784,
|
| 644 |
+
"grad_norm": 2.168471097946167,
|
| 645 |
+
"learning_rate": 1.4037265323518223e-05,
|
| 646 |
+
"loss": 0.133,
|
| 647 |
+
"step": 2500
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 15.434108527131784,
|
| 651 |
+
"eval_accuracy": 0.9547715965626413,
|
| 652 |
+
"eval_f1": 0.9545783302992185,
|
| 653 |
+
"eval_loss": 0.11651919037103653,
|
| 654 |
+
"eval_precision": 0.9549254114560721,
|
| 655 |
+
"eval_recall": 0.9547715965626413,
|
| 656 |
+
"eval_runtime": 31.9187,
|
| 657 |
+
"eval_samples_per_second": 69.27,
|
| 658 |
+
"eval_steps_per_second": 8.678,
|
| 659 |
+
"step": 2500
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 15.744186046511627,
|
| 663 |
+
"grad_norm": 4.06069278717041,
|
| 664 |
+
"learning_rate": 1.375996072327573e-05,
|
| 665 |
+
"loss": 0.117,
|
| 666 |
+
"step": 2550
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 16.049612403100774,
|
| 670 |
+
"grad_norm": 0.5793169736862183,
|
| 671 |
+
"learning_rate": 1.3479246431631214e-05,
|
| 672 |
+
"loss": 0.1258,
|
| 673 |
+
"step": 2600
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 16.049612403100774,
|
| 677 |
+
"eval_accuracy": 0.9479873360470376,
|
| 678 |
+
"eval_f1": 0.9473924145380525,
|
| 679 |
+
"eval_loss": 0.1302204728126526,
|
| 680 |
+
"eval_precision": 0.9500128100803557,
|
| 681 |
+
"eval_recall": 0.9479873360470376,
|
| 682 |
+
"eval_runtime": 32.0659,
|
| 683 |
+
"eval_samples_per_second": 68.952,
|
| 684 |
+
"eval_steps_per_second": 8.638,
|
| 685 |
+
"step": 2600
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 16.35968992248062,
|
| 689 |
+
"grad_norm": 3.8317296504974365,
|
| 690 |
+
"learning_rate": 1.3195377012176449e-05,
|
| 691 |
+
"loss": 0.1479,
|
| 692 |
+
"step": 2650
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 16.669767441860465,
|
| 696 |
+
"grad_norm": 3.2260377407073975,
|
| 697 |
+
"learning_rate": 1.2908609889706538e-05,
|
| 698 |
+
"loss": 0.115,
|
| 699 |
+
"step": 2700
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 16.669767441860465,
|
| 703 |
+
"eval_accuracy": 0.9534147444595206,
|
| 704 |
+
"eval_f1": 0.9536709713857593,
|
| 705 |
+
"eval_loss": 0.1319713443517685,
|
| 706 |
+
"eval_precision": 0.9552310659222302,
|
| 707 |
+
"eval_recall": 0.9534147444595206,
|
| 708 |
+
"eval_runtime": 31.9757,
|
| 709 |
+
"eval_samples_per_second": 69.146,
|
| 710 |
+
"eval_steps_per_second": 8.663,
|
| 711 |
+
"step": 2700
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 16.97984496124031,
|
| 715 |
+
"grad_norm": 1.9483188390731812,
|
| 716 |
+
"learning_rate": 1.2619205116776244e-05,
|
| 717 |
+
"loss": 0.1269,
|
| 718 |
+
"step": 2750
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 17.28527131782946,
|
| 722 |
+
"grad_norm": 4.348479270935059,
|
| 723 |
+
"learning_rate": 1.2327425137873315e-05,
|
| 724 |
+
"loss": 0.1134,
|
| 725 |
+
"step": 2800
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 17.28527131782946,
|
| 729 |
+
"eval_accuracy": 0.9552238805970149,
|
| 730 |
+
"eval_f1": 0.9548701351406266,
|
| 731 |
+
"eval_loss": 0.117097869515419,
|
| 732 |
+
"eval_precision": 0.9561949341655123,
|
| 733 |
+
"eval_recall": 0.9552238805970149,
|
| 734 |
+
"eval_runtime": 31.9852,
|
| 735 |
+
"eval_samples_per_second": 69.126,
|
| 736 |
+
"eval_steps_per_second": 8.66,
|
| 737 |
+
"step": 2800
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 17.5953488372093,
|
| 741 |
+
"grad_norm": 5.097657203674316,
|
| 742 |
+
"learning_rate": 1.2033534551422718e-05,
|
| 743 |
+
"loss": 0.1069,
|
| 744 |
+
"step": 2850
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 17.905426356589146,
|
| 748 |
+
"grad_norm": 2.5790622234344482,
|
| 749 |
+
"learning_rate": 1.173779986983762e-05,
|
| 750 |
+
"loss": 0.1069,
|
| 751 |
+
"step": 2900
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 17.905426356589146,
|
| 755 |
+
"eval_accuracy": 0.9597467209407508,
|
| 756 |
+
"eval_f1": 0.9595276897831222,
|
| 757 |
+
"eval_loss": 0.11270730942487717,
|
| 758 |
+
"eval_precision": 0.9601617456499858,
|
| 759 |
+
"eval_recall": 0.9597467209407508,
|
| 760 |
+
"eval_runtime": 32.0361,
|
| 761 |
+
"eval_samples_per_second": 69.016,
|
| 762 |
+
"eval_steps_per_second": 8.646,
|
| 763 |
+
"step": 2900
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 17.905426356589146,
|
| 767 |
+
"step": 2900,
|
| 768 |
+
"total_flos": 1.0232905043346113e+19,
|
| 769 |
+
"train_loss": 0.2441097328580659,
|
| 770 |
+
"train_runtime": 4508.9942,
|
| 771 |
+
"train_samples_per_second": 41.182,
|
| 772 |
+
"train_steps_per_second": 1.285
|
| 773 |
+
}
|
| 774 |
+
],
|
| 775 |
+
"logging_steps": 50,
|
| 776 |
+
"max_steps": 5796,
|
| 777 |
+
"num_input_tokens_seen": 0,
|
| 778 |
+
"num_train_epochs": 36,
|
| 779 |
+
"save_steps": 100,
|
| 780 |
+
"stateful_callbacks": {
|
| 781 |
+
"TrainerControl": {
|
| 782 |
+
"args": {
|
| 783 |
+
"should_epoch_stop": false,
|
| 784 |
+
"should_evaluate": false,
|
| 785 |
+
"should_log": false,
|
| 786 |
+
"should_save": true,
|
| 787 |
+
"should_training_stop": true
|
| 788 |
+
},
|
| 789 |
+
"attributes": {}
|
| 790 |
+
}
|
| 791 |
+
},
|
| 792 |
+
"total_flos": 1.0232905043346113e+19,
|
| 793 |
+
"train_batch_size": 8,
|
| 794 |
+
"trial_name": null,
|
| 795 |
+
"trial_params": null
|
| 796 |
+
}
|