End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 7.12,
|
| 3 |
+
"eval_accuracy": 0.7916666666666666,
|
| 4 |
+
"eval_loss": 0.6821911931037903,
|
| 5 |
+
"eval_runtime": 126.6935,
|
| 6 |
+
"eval_samples_per_second": 1.515,
|
| 7 |
+
"eval_steps_per_second": 0.189
|
| 8 |
+
}
|
runs/Jul28_17-15-50_518dd1f23e61/events.out.tfevents.1690577851.518dd1f23e61.2810.1
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:300d598a773ec7bb508861c9d43c95e5c49438c01b78bc60753c6dcb5fc5f9b0
|
| 3 |
+
size 734
|
test_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 7.12,
|
| 3 |
+
"eval_accuracy": 0.7916666666666666,
|
| 4 |
+
"eval_loss": 0.6821911931037903,
|
| 5 |
+
"eval_runtime": 126.6935,
|
| 6 |
+
"eval_samples_per_second": 1.515,
|
| 7 |
+
"eval_steps_per_second": 0.189
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1003 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.8085106382978723,
|
| 3 |
+
"best_model_checkpoint": "videomae-base-finetuned-movienet-finetuned-movienet-more/checkpoint-372",
|
| 4 |
+
"epoch": 7.120270270270271,
|
| 5 |
+
"global_step": 1480,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.01,
|
| 12 |
+
"learning_rate": 3.3783783783783788e-06,
|
| 13 |
+
"loss": 0.6452,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.01,
|
| 18 |
+
"learning_rate": 6.7567567567567575e-06,
|
| 19 |
+
"loss": 0.3741,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.02,
|
| 24 |
+
"learning_rate": 1.0135135135135136e-05,
|
| 25 |
+
"loss": 0.2853,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.03,
|
| 30 |
+
"learning_rate": 1.3513513513513515e-05,
|
| 31 |
+
"loss": 0.5599,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.03,
|
| 36 |
+
"learning_rate": 1.6891891891891892e-05,
|
| 37 |
+
"loss": 0.3713,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.04,
|
| 42 |
+
"learning_rate": 2.0270270270270273e-05,
|
| 43 |
+
"loss": 0.1326,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.05,
|
| 48 |
+
"learning_rate": 2.364864864864865e-05,
|
| 49 |
+
"loss": 0.1683,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.05,
|
| 54 |
+
"learning_rate": 2.702702702702703e-05,
|
| 55 |
+
"loss": 0.4612,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.06,
|
| 60 |
+
"learning_rate": 3.0405405405405407e-05,
|
| 61 |
+
"loss": 0.3596,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.07,
|
| 66 |
+
"learning_rate": 3.3783783783783784e-05,
|
| 67 |
+
"loss": 0.2449,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.07,
|
| 72 |
+
"learning_rate": 3.7162162162162165e-05,
|
| 73 |
+
"loss": 0.1317,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.08,
|
| 78 |
+
"learning_rate": 4.0540540540540545e-05,
|
| 79 |
+
"loss": 0.53,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.09,
|
| 84 |
+
"learning_rate": 4.391891891891892e-05,
|
| 85 |
+
"loss": 0.1933,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.09,
|
| 90 |
+
"learning_rate": 4.72972972972973e-05,
|
| 91 |
+
"loss": 0.3087,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.1,
|
| 96 |
+
"learning_rate": 4.9924924924924924e-05,
|
| 97 |
+
"loss": 0.365,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.11,
|
| 102 |
+
"learning_rate": 4.954954954954955e-05,
|
| 103 |
+
"loss": 0.3458,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.11,
|
| 108 |
+
"learning_rate": 4.9174174174174175e-05,
|
| 109 |
+
"loss": 0.4004,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.12,
|
| 114 |
+
"learning_rate": 4.87987987987988e-05,
|
| 115 |
+
"loss": 0.4977,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.13,
|
| 120 |
+
"eval_accuracy": 0.723404255319149,
|
| 121 |
+
"eval_loss": 0.8209818601608276,
|
| 122 |
+
"eval_runtime": 325.1637,
|
| 123 |
+
"eval_samples_per_second": 0.578,
|
| 124 |
+
"eval_steps_per_second": 0.074,
|
| 125 |
+
"step": 186
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 1.0,
|
| 129 |
+
"learning_rate": 4.8423423423423426e-05,
|
| 130 |
+
"loss": 0.31,
|
| 131 |
+
"step": 190
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 1.01,
|
| 135 |
+
"learning_rate": 4.804804804804805e-05,
|
| 136 |
+
"loss": 0.4281,
|
| 137 |
+
"step": 200
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.02,
|
| 141 |
+
"learning_rate": 4.767267267267268e-05,
|
| 142 |
+
"loss": 0.5228,
|
| 143 |
+
"step": 210
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 1.02,
|
| 147 |
+
"learning_rate": 4.72972972972973e-05,
|
| 148 |
+
"loss": 0.4479,
|
| 149 |
+
"step": 220
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 1.03,
|
| 153 |
+
"learning_rate": 4.692192192192192e-05,
|
| 154 |
+
"loss": 0.4205,
|
| 155 |
+
"step": 230
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 1.04,
|
| 159 |
+
"learning_rate": 4.654654654654655e-05,
|
| 160 |
+
"loss": 0.6442,
|
| 161 |
+
"step": 240
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 1.04,
|
| 165 |
+
"learning_rate": 4.617117117117117e-05,
|
| 166 |
+
"loss": 0.5665,
|
| 167 |
+
"step": 250
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 1.05,
|
| 171 |
+
"learning_rate": 4.57957957957958e-05,
|
| 172 |
+
"loss": 0.4637,
|
| 173 |
+
"step": 260
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 1.06,
|
| 177 |
+
"learning_rate": 4.5420420420420424e-05,
|
| 178 |
+
"loss": 0.2689,
|
| 179 |
+
"step": 270
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.06,
|
| 183 |
+
"learning_rate": 4.5045045045045046e-05,
|
| 184 |
+
"loss": 0.2813,
|
| 185 |
+
"step": 280
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.07,
|
| 189 |
+
"learning_rate": 4.4669669669669675e-05,
|
| 190 |
+
"loss": 0.5086,
|
| 191 |
+
"step": 290
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.08,
|
| 195 |
+
"learning_rate": 4.42942942942943e-05,
|
| 196 |
+
"loss": 0.2338,
|
| 197 |
+
"step": 300
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 1.08,
|
| 201 |
+
"learning_rate": 4.391891891891892e-05,
|
| 202 |
+
"loss": 0.4156,
|
| 203 |
+
"step": 310
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 1.09,
|
| 207 |
+
"learning_rate": 4.354354354354355e-05,
|
| 208 |
+
"loss": 0.4025,
|
| 209 |
+
"step": 320
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 1.1,
|
| 213 |
+
"learning_rate": 4.316816816816817e-05,
|
| 214 |
+
"loss": 0.3151,
|
| 215 |
+
"step": 330
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 1.1,
|
| 219 |
+
"learning_rate": 4.27927927927928e-05,
|
| 220 |
+
"loss": 0.4204,
|
| 221 |
+
"step": 340
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.11,
|
| 225 |
+
"learning_rate": 4.241741741741742e-05,
|
| 226 |
+
"loss": 0.4158,
|
| 227 |
+
"step": 350
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 1.12,
|
| 231 |
+
"learning_rate": 4.204204204204204e-05,
|
| 232 |
+
"loss": 0.3485,
|
| 233 |
+
"step": 360
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.12,
|
| 237 |
+
"learning_rate": 4.166666666666667e-05,
|
| 238 |
+
"loss": 0.3974,
|
| 239 |
+
"step": 370
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.13,
|
| 243 |
+
"eval_accuracy": 0.8085106382978723,
|
| 244 |
+
"eval_loss": 0.632224977016449,
|
| 245 |
+
"eval_runtime": 167.3633,
|
| 246 |
+
"eval_samples_per_second": 1.123,
|
| 247 |
+
"eval_steps_per_second": 0.143,
|
| 248 |
+
"step": 372
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 2.01,
|
| 252 |
+
"learning_rate": 4.1291291291291294e-05,
|
| 253 |
+
"loss": 0.2036,
|
| 254 |
+
"step": 380
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 2.01,
|
| 258 |
+
"learning_rate": 4.091591591591592e-05,
|
| 259 |
+
"loss": 0.2391,
|
| 260 |
+
"step": 390
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 2.02,
|
| 264 |
+
"learning_rate": 4.0540540540540545e-05,
|
| 265 |
+
"loss": 0.2951,
|
| 266 |
+
"step": 400
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 2.03,
|
| 270 |
+
"learning_rate": 4.016516516516517e-05,
|
| 271 |
+
"loss": 0.3131,
|
| 272 |
+
"step": 410
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.03,
|
| 276 |
+
"learning_rate": 3.9789789789789796e-05,
|
| 277 |
+
"loss": 0.3097,
|
| 278 |
+
"step": 420
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 2.04,
|
| 282 |
+
"learning_rate": 3.941441441441442e-05,
|
| 283 |
+
"loss": 0.1422,
|
| 284 |
+
"step": 430
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 2.05,
|
| 288 |
+
"learning_rate": 3.903903903903904e-05,
|
| 289 |
+
"loss": 0.0668,
|
| 290 |
+
"step": 440
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 2.05,
|
| 294 |
+
"learning_rate": 3.866366366366366e-05,
|
| 295 |
+
"loss": 0.2289,
|
| 296 |
+
"step": 450
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 2.06,
|
| 300 |
+
"learning_rate": 3.8288288288288285e-05,
|
| 301 |
+
"loss": 0.1771,
|
| 302 |
+
"step": 460
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 2.07,
|
| 306 |
+
"learning_rate": 3.7912912912912914e-05,
|
| 307 |
+
"loss": 0.1779,
|
| 308 |
+
"step": 470
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 2.07,
|
| 312 |
+
"learning_rate": 3.7537537537537536e-05,
|
| 313 |
+
"loss": 0.6673,
|
| 314 |
+
"step": 480
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.08,
|
| 318 |
+
"learning_rate": 3.7162162162162165e-05,
|
| 319 |
+
"loss": 0.2319,
|
| 320 |
+
"step": 490
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 2.09,
|
| 324 |
+
"learning_rate": 3.678678678678679e-05,
|
| 325 |
+
"loss": 0.157,
|
| 326 |
+
"step": 500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 2.09,
|
| 330 |
+
"learning_rate": 3.641141141141141e-05,
|
| 331 |
+
"loss": 0.2388,
|
| 332 |
+
"step": 510
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 2.1,
|
| 336 |
+
"learning_rate": 3.603603603603604e-05,
|
| 337 |
+
"loss": 0.4595,
|
| 338 |
+
"step": 520
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 2.11,
|
| 342 |
+
"learning_rate": 3.566066066066066e-05,
|
| 343 |
+
"loss": 0.2776,
|
| 344 |
+
"step": 530
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 2.11,
|
| 348 |
+
"learning_rate": 3.528528528528528e-05,
|
| 349 |
+
"loss": 0.2687,
|
| 350 |
+
"step": 540
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 2.12,
|
| 354 |
+
"learning_rate": 3.490990990990991e-05,
|
| 355 |
+
"loss": 0.1101,
|
| 356 |
+
"step": 550
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.13,
|
| 360 |
+
"eval_accuracy": 0.75,
|
| 361 |
+
"eval_loss": 1.0521998405456543,
|
| 362 |
+
"eval_runtime": 162.7827,
|
| 363 |
+
"eval_samples_per_second": 1.155,
|
| 364 |
+
"eval_steps_per_second": 0.147,
|
| 365 |
+
"step": 558
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 3.0,
|
| 369 |
+
"learning_rate": 3.453453453453453e-05,
|
| 370 |
+
"loss": 0.1566,
|
| 371 |
+
"step": 560
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 3.01,
|
| 375 |
+
"learning_rate": 3.415915915915916e-05,
|
| 376 |
+
"loss": 0.2506,
|
| 377 |
+
"step": 570
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 3.01,
|
| 381 |
+
"learning_rate": 3.3783783783783784e-05,
|
| 382 |
+
"loss": 0.0709,
|
| 383 |
+
"step": 580
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 3.02,
|
| 387 |
+
"learning_rate": 3.3408408408408406e-05,
|
| 388 |
+
"loss": 0.1136,
|
| 389 |
+
"step": 590
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 3.03,
|
| 393 |
+
"learning_rate": 3.3033033033033035e-05,
|
| 394 |
+
"loss": 0.3604,
|
| 395 |
+
"step": 600
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 3.04,
|
| 399 |
+
"learning_rate": 3.265765765765766e-05,
|
| 400 |
+
"loss": 0.0949,
|
| 401 |
+
"step": 610
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 3.04,
|
| 405 |
+
"learning_rate": 3.2282282282282286e-05,
|
| 406 |
+
"loss": 0.038,
|
| 407 |
+
"step": 620
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 3.05,
|
| 411 |
+
"learning_rate": 3.190690690690691e-05,
|
| 412 |
+
"loss": 0.0487,
|
| 413 |
+
"step": 630
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 3.06,
|
| 417 |
+
"learning_rate": 3.153153153153153e-05,
|
| 418 |
+
"loss": 0.1361,
|
| 419 |
+
"step": 640
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 3.06,
|
| 423 |
+
"learning_rate": 3.115615615615616e-05,
|
| 424 |
+
"loss": 0.2508,
|
| 425 |
+
"step": 650
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 3.07,
|
| 429 |
+
"learning_rate": 3.078078078078078e-05,
|
| 430 |
+
"loss": 0.0955,
|
| 431 |
+
"step": 660
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 3.08,
|
| 435 |
+
"learning_rate": 3.0405405405405407e-05,
|
| 436 |
+
"loss": 0.2185,
|
| 437 |
+
"step": 670
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 3.08,
|
| 441 |
+
"learning_rate": 3.0030030030030033e-05,
|
| 442 |
+
"loss": 0.3311,
|
| 443 |
+
"step": 680
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 3.09,
|
| 447 |
+
"learning_rate": 2.9654654654654658e-05,
|
| 448 |
+
"loss": 0.1093,
|
| 449 |
+
"step": 690
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 3.1,
|
| 453 |
+
"learning_rate": 2.927927927927928e-05,
|
| 454 |
+
"loss": 0.0554,
|
| 455 |
+
"step": 700
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 3.1,
|
| 459 |
+
"learning_rate": 2.8903903903903906e-05,
|
| 460 |
+
"loss": 0.084,
|
| 461 |
+
"step": 710
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 3.11,
|
| 465 |
+
"learning_rate": 2.852852852852853e-05,
|
| 466 |
+
"loss": 0.1206,
|
| 467 |
+
"step": 720
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 3.12,
|
| 471 |
+
"learning_rate": 2.8153153153153157e-05,
|
| 472 |
+
"loss": 0.2526,
|
| 473 |
+
"step": 730
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 3.12,
|
| 477 |
+
"learning_rate": 2.777777777777778e-05,
|
| 478 |
+
"loss": 0.1618,
|
| 479 |
+
"step": 740
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 3.13,
|
| 483 |
+
"eval_accuracy": 0.7872340425531915,
|
| 484 |
+
"eval_loss": 0.9184951782226562,
|
| 485 |
+
"eval_runtime": 122.6151,
|
| 486 |
+
"eval_samples_per_second": 1.533,
|
| 487 |
+
"eval_steps_per_second": 0.196,
|
| 488 |
+
"step": 744
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 4.0,
|
| 492 |
+
"learning_rate": 2.7402402402402405e-05,
|
| 493 |
+
"loss": 0.0652,
|
| 494 |
+
"step": 750
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 4.01,
|
| 498 |
+
"learning_rate": 2.702702702702703e-05,
|
| 499 |
+
"loss": 0.027,
|
| 500 |
+
"step": 760
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 4.02,
|
| 504 |
+
"learning_rate": 2.6651651651651656e-05,
|
| 505 |
+
"loss": 0.0875,
|
| 506 |
+
"step": 770
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 4.02,
|
| 510 |
+
"learning_rate": 2.6276276276276278e-05,
|
| 511 |
+
"loss": 0.2058,
|
| 512 |
+
"step": 780
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 4.03,
|
| 516 |
+
"learning_rate": 2.5900900900900903e-05,
|
| 517 |
+
"loss": 0.0566,
|
| 518 |
+
"step": 790
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 4.04,
|
| 522 |
+
"learning_rate": 2.552552552552553e-05,
|
| 523 |
+
"loss": 0.082,
|
| 524 |
+
"step": 800
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 4.04,
|
| 528 |
+
"learning_rate": 2.5150150150150154e-05,
|
| 529 |
+
"loss": 0.0835,
|
| 530 |
+
"step": 810
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 4.05,
|
| 534 |
+
"learning_rate": 2.4774774774774777e-05,
|
| 535 |
+
"loss": 0.0831,
|
| 536 |
+
"step": 820
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 4.06,
|
| 540 |
+
"learning_rate": 2.43993993993994e-05,
|
| 541 |
+
"loss": 0.0566,
|
| 542 |
+
"step": 830
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 4.06,
|
| 546 |
+
"learning_rate": 2.4024024024024024e-05,
|
| 547 |
+
"loss": 0.053,
|
| 548 |
+
"step": 840
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 4.07,
|
| 552 |
+
"learning_rate": 2.364864864864865e-05,
|
| 553 |
+
"loss": 0.0415,
|
| 554 |
+
"step": 850
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 4.08,
|
| 558 |
+
"learning_rate": 2.3273273273273275e-05,
|
| 559 |
+
"loss": 0.072,
|
| 560 |
+
"step": 860
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 4.09,
|
| 564 |
+
"learning_rate": 2.28978978978979e-05,
|
| 565 |
+
"loss": 0.1109,
|
| 566 |
+
"step": 870
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 4.09,
|
| 570 |
+
"learning_rate": 2.2522522522522523e-05,
|
| 571 |
+
"loss": 0.1355,
|
| 572 |
+
"step": 880
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 4.1,
|
| 576 |
+
"learning_rate": 2.214714714714715e-05,
|
| 577 |
+
"loss": 0.0971,
|
| 578 |
+
"step": 890
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 4.11,
|
| 582 |
+
"learning_rate": 2.1771771771771774e-05,
|
| 583 |
+
"loss": 0.1569,
|
| 584 |
+
"step": 900
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 4.11,
|
| 588 |
+
"learning_rate": 2.13963963963964e-05,
|
| 589 |
+
"loss": 0.0111,
|
| 590 |
+
"step": 910
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 4.12,
|
| 594 |
+
"learning_rate": 2.102102102102102e-05,
|
| 595 |
+
"loss": 0.0249,
|
| 596 |
+
"step": 920
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 4.13,
|
| 600 |
+
"learning_rate": 2.0645645645645647e-05,
|
| 601 |
+
"loss": 0.0052,
|
| 602 |
+
"step": 930
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 4.13,
|
| 606 |
+
"eval_accuracy": 0.8085106382978723,
|
| 607 |
+
"eval_loss": 0.8585702776908875,
|
| 608 |
+
"eval_runtime": 123.8694,
|
| 609 |
+
"eval_samples_per_second": 1.518,
|
| 610 |
+
"eval_steps_per_second": 0.194,
|
| 611 |
+
"step": 930
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 5.01,
|
| 615 |
+
"learning_rate": 2.0270270270270273e-05,
|
| 616 |
+
"loss": 0.008,
|
| 617 |
+
"step": 940
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 5.01,
|
| 621 |
+
"learning_rate": 1.9894894894894898e-05,
|
| 622 |
+
"loss": 0.0054,
|
| 623 |
+
"step": 950
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 5.02,
|
| 627 |
+
"learning_rate": 1.951951951951952e-05,
|
| 628 |
+
"loss": 0.0067,
|
| 629 |
+
"step": 960
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 5.03,
|
| 633 |
+
"learning_rate": 1.9144144144144142e-05,
|
| 634 |
+
"loss": 0.0022,
|
| 635 |
+
"step": 970
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 5.03,
|
| 639 |
+
"learning_rate": 1.8768768768768768e-05,
|
| 640 |
+
"loss": 0.0504,
|
| 641 |
+
"step": 980
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 5.04,
|
| 645 |
+
"learning_rate": 1.8393393393393393e-05,
|
| 646 |
+
"loss": 0.001,
|
| 647 |
+
"step": 990
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 5.05,
|
| 651 |
+
"learning_rate": 1.801801801801802e-05,
|
| 652 |
+
"loss": 0.0017,
|
| 653 |
+
"step": 1000
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 5.05,
|
| 657 |
+
"learning_rate": 1.764264264264264e-05,
|
| 658 |
+
"loss": 0.0025,
|
| 659 |
+
"step": 1010
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 5.06,
|
| 663 |
+
"learning_rate": 1.7267267267267267e-05,
|
| 664 |
+
"loss": 0.1008,
|
| 665 |
+
"step": 1020
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 5.07,
|
| 669 |
+
"learning_rate": 1.6891891891891892e-05,
|
| 670 |
+
"loss": 0.0033,
|
| 671 |
+
"step": 1030
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 5.07,
|
| 675 |
+
"learning_rate": 1.6516516516516518e-05,
|
| 676 |
+
"loss": 0.0022,
|
| 677 |
+
"step": 1040
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 5.08,
|
| 681 |
+
"learning_rate": 1.6141141141141143e-05,
|
| 682 |
+
"loss": 0.0122,
|
| 683 |
+
"step": 1050
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 5.09,
|
| 687 |
+
"learning_rate": 1.5765765765765765e-05,
|
| 688 |
+
"loss": 0.0615,
|
| 689 |
+
"step": 1060
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 5.09,
|
| 693 |
+
"learning_rate": 1.539039039039039e-05,
|
| 694 |
+
"loss": 0.1198,
|
| 695 |
+
"step": 1070
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 5.1,
|
| 699 |
+
"learning_rate": 1.5015015015015016e-05,
|
| 700 |
+
"loss": 0.0386,
|
| 701 |
+
"step": 1080
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 5.11,
|
| 705 |
+
"learning_rate": 1.463963963963964e-05,
|
| 706 |
+
"loss": 0.0491,
|
| 707 |
+
"step": 1090
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 5.11,
|
| 711 |
+
"learning_rate": 1.4264264264264266e-05,
|
| 712 |
+
"loss": 0.0265,
|
| 713 |
+
"step": 1100
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 5.12,
|
| 717 |
+
"learning_rate": 1.388888888888889e-05,
|
| 718 |
+
"loss": 0.0013,
|
| 719 |
+
"step": 1110
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 5.13,
|
| 723 |
+
"eval_accuracy": 0.7925531914893617,
|
| 724 |
+
"eval_loss": 1.0028539896011353,
|
| 725 |
+
"eval_runtime": 127.0573,
|
| 726 |
+
"eval_samples_per_second": 1.48,
|
| 727 |
+
"eval_steps_per_second": 0.189,
|
| 728 |
+
"step": 1116
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 6.0,
|
| 732 |
+
"learning_rate": 1.3513513513513515e-05,
|
| 733 |
+
"loss": 0.0228,
|
| 734 |
+
"step": 1120
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 6.01,
|
| 738 |
+
"learning_rate": 1.3138138138138139e-05,
|
| 739 |
+
"loss": 0.0011,
|
| 740 |
+
"step": 1130
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 6.02,
|
| 744 |
+
"learning_rate": 1.2762762762762764e-05,
|
| 745 |
+
"loss": 0.0247,
|
| 746 |
+
"step": 1140
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 6.02,
|
| 750 |
+
"learning_rate": 1.2387387387387388e-05,
|
| 751 |
+
"loss": 0.0048,
|
| 752 |
+
"step": 1150
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 6.03,
|
| 756 |
+
"learning_rate": 1.2012012012012012e-05,
|
| 757 |
+
"loss": 0.0009,
|
| 758 |
+
"step": 1160
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 6.04,
|
| 762 |
+
"learning_rate": 1.1636636636636638e-05,
|
| 763 |
+
"loss": 0.0065,
|
| 764 |
+
"step": 1170
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 6.04,
|
| 768 |
+
"learning_rate": 1.1261261261261261e-05,
|
| 769 |
+
"loss": 0.0012,
|
| 770 |
+
"step": 1180
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 6.05,
|
| 774 |
+
"learning_rate": 1.0885885885885887e-05,
|
| 775 |
+
"loss": 0.0036,
|
| 776 |
+
"step": 1190
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 6.06,
|
| 780 |
+
"learning_rate": 1.051051051051051e-05,
|
| 781 |
+
"loss": 0.0009,
|
| 782 |
+
"step": 1200
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 6.06,
|
| 786 |
+
"learning_rate": 1.0135135135135136e-05,
|
| 787 |
+
"loss": 0.0115,
|
| 788 |
+
"step": 1210
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 6.07,
|
| 792 |
+
"learning_rate": 9.75975975975976e-06,
|
| 793 |
+
"loss": 0.011,
|
| 794 |
+
"step": 1220
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 6.08,
|
| 798 |
+
"learning_rate": 9.384384384384384e-06,
|
| 799 |
+
"loss": 0.0318,
|
| 800 |
+
"step": 1230
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 6.08,
|
| 804 |
+
"learning_rate": 9.00900900900901e-06,
|
| 805 |
+
"loss": 0.0025,
|
| 806 |
+
"step": 1240
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 6.09,
|
| 810 |
+
"learning_rate": 8.633633633633633e-06,
|
| 811 |
+
"loss": 0.0373,
|
| 812 |
+
"step": 1250
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 6.1,
|
| 816 |
+
"learning_rate": 8.258258258258259e-06,
|
| 817 |
+
"loss": 0.0651,
|
| 818 |
+
"step": 1260
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 6.1,
|
| 822 |
+
"learning_rate": 7.882882882882883e-06,
|
| 823 |
+
"loss": 0.0052,
|
| 824 |
+
"step": 1270
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 6.11,
|
| 828 |
+
"learning_rate": 7.507507507507508e-06,
|
| 829 |
+
"loss": 0.0008,
|
| 830 |
+
"step": 1280
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 6.12,
|
| 834 |
+
"learning_rate": 7.132132132132133e-06,
|
| 835 |
+
"loss": 0.0013,
|
| 836 |
+
"step": 1290
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 6.12,
|
| 840 |
+
"learning_rate": 6.7567567567567575e-06,
|
| 841 |
+
"loss": 0.0044,
|
| 842 |
+
"step": 1300
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 6.13,
|
| 846 |
+
"eval_accuracy": 0.7872340425531915,
|
| 847 |
+
"eval_loss": 1.0399647951126099,
|
| 848 |
+
"eval_runtime": 160.3025,
|
| 849 |
+
"eval_samples_per_second": 1.173,
|
| 850 |
+
"eval_steps_per_second": 0.15,
|
| 851 |
+
"step": 1302
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 7.01,
|
| 855 |
+
"learning_rate": 6.381381381381382e-06,
|
| 856 |
+
"loss": 0.0009,
|
| 857 |
+
"step": 1310
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 7.01,
|
| 861 |
+
"learning_rate": 6.006006006006006e-06,
|
| 862 |
+
"loss": 0.0016,
|
| 863 |
+
"step": 1320
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 7.02,
|
| 867 |
+
"learning_rate": 5.630630630630631e-06,
|
| 868 |
+
"loss": 0.0008,
|
| 869 |
+
"step": 1330
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 7.03,
|
| 873 |
+
"learning_rate": 5.255255255255255e-06,
|
| 874 |
+
"loss": 0.0008,
|
| 875 |
+
"step": 1340
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 7.03,
|
| 879 |
+
"learning_rate": 4.87987987987988e-06,
|
| 880 |
+
"loss": 0.0009,
|
| 881 |
+
"step": 1350
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 7.04,
|
| 885 |
+
"learning_rate": 4.504504504504505e-06,
|
| 886 |
+
"loss": 0.0007,
|
| 887 |
+
"step": 1360
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 7.05,
|
| 891 |
+
"learning_rate": 4.129129129129129e-06,
|
| 892 |
+
"loss": 0.001,
|
| 893 |
+
"step": 1370
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 7.05,
|
| 897 |
+
"learning_rate": 3.753753753753754e-06,
|
| 898 |
+
"loss": 0.0008,
|
| 899 |
+
"step": 1380
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 7.06,
|
| 903 |
+
"learning_rate": 3.3783783783783788e-06,
|
| 904 |
+
"loss": 0.0007,
|
| 905 |
+
"step": 1390
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 7.07,
|
| 909 |
+
"learning_rate": 3.003003003003003e-06,
|
| 910 |
+
"loss": 0.001,
|
| 911 |
+
"step": 1400
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 7.07,
|
| 915 |
+
"learning_rate": 2.6276276276276277e-06,
|
| 916 |
+
"loss": 0.0007,
|
| 917 |
+
"step": 1410
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 7.08,
|
| 921 |
+
"learning_rate": 2.2522522522522524e-06,
|
| 922 |
+
"loss": 0.016,
|
| 923 |
+
"step": 1420
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 7.09,
|
| 927 |
+
"learning_rate": 1.876876876876877e-06,
|
| 928 |
+
"loss": 0.0008,
|
| 929 |
+
"step": 1430
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 7.09,
|
| 933 |
+
"learning_rate": 1.5015015015015015e-06,
|
| 934 |
+
"loss": 0.0023,
|
| 935 |
+
"step": 1440
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 7.1,
|
| 939 |
+
"learning_rate": 1.1261261261261262e-06,
|
| 940 |
+
"loss": 0.0007,
|
| 941 |
+
"step": 1450
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 7.11,
|
| 945 |
+
"learning_rate": 7.507507507507508e-07,
|
| 946 |
+
"loss": 0.0007,
|
| 947 |
+
"step": 1460
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 7.11,
|
| 951 |
+
"learning_rate": 3.753753753753754e-07,
|
| 952 |
+
"loss": 0.0007,
|
| 953 |
+
"step": 1470
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 7.12,
|
| 957 |
+
"learning_rate": 0.0,
|
| 958 |
+
"loss": 0.0008,
|
| 959 |
+
"step": 1480
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 7.12,
|
| 963 |
+
"eval_accuracy": 0.7978723404255319,
|
| 964 |
+
"eval_loss": 1.0017523765563965,
|
| 965 |
+
"eval_runtime": 123.2439,
|
| 966 |
+
"eval_samples_per_second": 1.525,
|
| 967 |
+
"eval_steps_per_second": 0.195,
|
| 968 |
+
"step": 1480
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 7.12,
|
| 972 |
+
"step": 1480,
|
| 973 |
+
"total_flos": 1.4692740659720847e+19,
|
| 974 |
+
"train_loss": 0.1616095771478187,
|
| 975 |
+
"train_runtime": 12949.9285,
|
| 976 |
+
"train_samples_per_second": 0.914,
|
| 977 |
+
"train_steps_per_second": 0.114
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 7.12,
|
| 981 |
+
"eval_accuracy": 0.7916666666666666,
|
| 982 |
+
"eval_loss": 0.6821911931037903,
|
| 983 |
+
"eval_runtime": 285.028,
|
| 984 |
+
"eval_samples_per_second": 0.674,
|
| 985 |
+
"eval_steps_per_second": 0.084,
|
| 986 |
+
"step": 1480
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 7.12,
|
| 990 |
+
"eval_accuracy": 0.7916666666666666,
|
| 991 |
+
"eval_loss": 0.6821911931037903,
|
| 992 |
+
"eval_runtime": 126.6935,
|
| 993 |
+
"eval_samples_per_second": 1.515,
|
| 994 |
+
"eval_steps_per_second": 0.189,
|
| 995 |
+
"step": 1480
|
| 996 |
+
}
|
| 997 |
+
],
|
| 998 |
+
"max_steps": 1480,
|
| 999 |
+
"num_train_epochs": 9223372036854775807,
|
| 1000 |
+
"total_flos": 1.4692740659720847e+19,
|
| 1001 |
+
"trial_name": null,
|
| 1002 |
+
"trial_params": null
|
| 1003 |
+
}
|