Commit
•
848070f
1
Parent(s):
234473f
Upload 9 files
Browse filesacc on test 0.965
- checkpoint-3768/config.json +36 -0
- checkpoint-3768/optimizer.pt +3 -0
- checkpoint-3768/preprocessor_config.json +22 -0
- checkpoint-3768/pytorch_model.bin +3 -0
- checkpoint-3768/rng_state.pth +3 -0
- checkpoint-3768/scaler.pt +3 -0
- checkpoint-3768/scheduler.pt +3 -0
- checkpoint-3768/trainer_state.json +2488 -0
- checkpoint-3768/training_args.bin +3 -0
checkpoint-3768/config.json
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/vit-base-patch16-224",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"ViTForImageClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.0,
|
| 7 |
+
"encoder_stride": 16,
|
| 8 |
+
"hidden_act": "gelu",
|
| 9 |
+
"hidden_dropout_prob": 0.0,
|
| 10 |
+
"hidden_size": 768,
|
| 11 |
+
"id2label": {
|
| 12 |
+
"0": "adenocarcinoma",
|
| 13 |
+
"1": "large.cell",
|
| 14 |
+
"2": "normal",
|
| 15 |
+
"3": "squamous.cell"
|
| 16 |
+
},
|
| 17 |
+
"image_size": 224,
|
| 18 |
+
"initializer_range": 0.02,
|
| 19 |
+
"intermediate_size": 3072,
|
| 20 |
+
"label2id": {
|
| 21 |
+
"adenocarcinoma": 0,
|
| 22 |
+
"large.cell": 1,
|
| 23 |
+
"normal": 2,
|
| 24 |
+
"squamous.cell": 3
|
| 25 |
+
},
|
| 26 |
+
"layer_norm_eps": 1e-12,
|
| 27 |
+
"model_type": "vit",
|
| 28 |
+
"num_attention_heads": 12,
|
| 29 |
+
"num_channels": 3,
|
| 30 |
+
"num_hidden_layers": 12,
|
| 31 |
+
"patch_size": 16,
|
| 32 |
+
"problem_type": "single_label_classification",
|
| 33 |
+
"qkv_bias": true,
|
| 34 |
+
"torch_dtype": "float32",
|
| 35 |
+
"transformers_version": "4.27.4"
|
| 36 |
+
}
|
checkpoint-3768/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f01b68c103e7f2cfa5932a5ade859cf682988374cf2abbd0a31e5d0abf2f5c54
|
| 3 |
+
size 686531205
|
checkpoint-3768/preprocessor_config.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"do_rescale": true,
|
| 4 |
+
"do_resize": true,
|
| 5 |
+
"image_mean": [
|
| 6 |
+
0.5,
|
| 7 |
+
0.5,
|
| 8 |
+
0.5
|
| 9 |
+
],
|
| 10 |
+
"image_processor_type": "ViTImageProcessor",
|
| 11 |
+
"image_std": [
|
| 12 |
+
0.5,
|
| 13 |
+
0.5,
|
| 14 |
+
0.5
|
| 15 |
+
],
|
| 16 |
+
"resample": 2,
|
| 17 |
+
"rescale_factor": 0.00392156862745098,
|
| 18 |
+
"size": {
|
| 19 |
+
"height": 224,
|
| 20 |
+
"width": 224
|
| 21 |
+
}
|
| 22 |
+
}
|
checkpoint-3768/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:60f2e6e535dbd8e8ba1d9787abbdfe5263b67bbb62ae7417641092a3ef826d7b
|
| 3 |
+
size 343274861
|
checkpoint-3768/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f038abe301d6561297f4a7beef62d40fc02ba8fdadec15dd550eb9366cdce0da
|
| 3 |
+
size 14575
|
checkpoint-3768/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:aba265632a0ddb5cc5055099d68a327ca45a7986ab77a88a5645c44e9bed9416
|
| 3 |
+
size 557
|
checkpoint-3768/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:78e6fd6981969e4f36445c784b3bf789e21d9c7fc0206141b3d5642a765a65e7
|
| 3 |
+
size 627
|
checkpoint-3768/trainer_state.json
ADDED
|
@@ -0,0 +1,2488 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.97,
|
| 3 |
+
"best_model_checkpoint": "./vit-LungCancerAUG10000img/checkpoint-3768",
|
| 4 |
+
"epoch": 24.0,
|
| 5 |
+
"global_step": 3768,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.06,
|
| 12 |
+
"learning_rate": 4.2462845010615706e-08,
|
| 13 |
+
"loss": 1.4586,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.13,
|
| 18 |
+
"learning_rate": 8.492569002123141e-08,
|
| 19 |
+
"loss": 1.4488,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.19,
|
| 24 |
+
"learning_rate": 1.2738853503184713e-07,
|
| 25 |
+
"loss": 1.4723,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.25,
|
| 30 |
+
"learning_rate": 1.6985138004246283e-07,
|
| 31 |
+
"loss": 1.4157,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.32,
|
| 36 |
+
"learning_rate": 2.1231422505307855e-07,
|
| 37 |
+
"loss": 1.4137,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.38,
|
| 42 |
+
"learning_rate": 2.5477707006369425e-07,
|
| 43 |
+
"loss": 1.3989,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.45,
|
| 48 |
+
"learning_rate": 2.9723991507431e-07,
|
| 49 |
+
"loss": 1.4097,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.51,
|
| 54 |
+
"learning_rate": 3.3970276008492565e-07,
|
| 55 |
+
"loss": 1.4044,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.57,
|
| 60 |
+
"learning_rate": 3.8216560509554143e-07,
|
| 61 |
+
"loss": 1.4031,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.64,
|
| 66 |
+
"learning_rate": 4.246284501061571e-07,
|
| 67 |
+
"loss": 1.3575,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.7,
|
| 72 |
+
"learning_rate": 4.6709129511677283e-07,
|
| 73 |
+
"loss": 1.3662,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.76,
|
| 78 |
+
"learning_rate": 5.095541401273885e-07,
|
| 79 |
+
"loss": 1.3338,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.83,
|
| 84 |
+
"learning_rate": 5.520169851380042e-07,
|
| 85 |
+
"loss": 1.3164,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.89,
|
| 90 |
+
"learning_rate": 5.9447983014862e-07,
|
| 91 |
+
"loss": 1.3017,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.96,
|
| 96 |
+
"learning_rate": 6.369426751592356e-07,
|
| 97 |
+
"loss": 1.2903,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.0,
|
| 102 |
+
"eval_accuracy": 0.445,
|
| 103 |
+
"eval_loss": 1.260408878326416,
|
| 104 |
+
"eval_runtime": 129.0129,
|
| 105 |
+
"eval_samples_per_second": 3.1,
|
| 106 |
+
"eval_steps_per_second": 0.388,
|
| 107 |
+
"step": 157
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.02,
|
| 111 |
+
"learning_rate": 6.794055201698513e-07,
|
| 112 |
+
"loss": 1.2839,
|
| 113 |
+
"step": 160
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 1.08,
|
| 117 |
+
"learning_rate": 7.21868365180467e-07,
|
| 118 |
+
"loss": 1.226,
|
| 119 |
+
"step": 170
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.15,
|
| 123 |
+
"learning_rate": 7.643312101910829e-07,
|
| 124 |
+
"loss": 1.2273,
|
| 125 |
+
"step": 180
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 1.21,
|
| 129 |
+
"learning_rate": 8.067940552016985e-07,
|
| 130 |
+
"loss": 1.2059,
|
| 131 |
+
"step": 190
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 1.27,
|
| 135 |
+
"learning_rate": 8.492569002123142e-07,
|
| 136 |
+
"loss": 1.1669,
|
| 137 |
+
"step": 200
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.34,
|
| 141 |
+
"learning_rate": 8.917197452229299e-07,
|
| 142 |
+
"loss": 1.1302,
|
| 143 |
+
"step": 210
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 1.4,
|
| 147 |
+
"learning_rate": 9.341825902335457e-07,
|
| 148 |
+
"loss": 1.1348,
|
| 149 |
+
"step": 220
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 1.46,
|
| 153 |
+
"learning_rate": 9.766454352441613e-07,
|
| 154 |
+
"loss": 1.0836,
|
| 155 |
+
"step": 230
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 1.53,
|
| 159 |
+
"learning_rate": 1.019108280254777e-06,
|
| 160 |
+
"loss": 1.0167,
|
| 161 |
+
"step": 240
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 1.59,
|
| 165 |
+
"learning_rate": 1.0615711252653927e-06,
|
| 166 |
+
"loss": 1.0234,
|
| 167 |
+
"step": 250
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 1.66,
|
| 171 |
+
"learning_rate": 1.1040339702760084e-06,
|
| 172 |
+
"loss": 1.0272,
|
| 173 |
+
"step": 260
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 1.72,
|
| 177 |
+
"learning_rate": 1.146496815286624e-06,
|
| 178 |
+
"loss": 0.9718,
|
| 179 |
+
"step": 270
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.78,
|
| 183 |
+
"learning_rate": 1.18895966029724e-06,
|
| 184 |
+
"loss": 0.9686,
|
| 185 |
+
"step": 280
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.85,
|
| 189 |
+
"learning_rate": 1.2314225053078556e-06,
|
| 190 |
+
"loss": 0.9369,
|
| 191 |
+
"step": 290
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.91,
|
| 195 |
+
"learning_rate": 1.2738853503184713e-06,
|
| 196 |
+
"loss": 0.9313,
|
| 197 |
+
"step": 300
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 1.97,
|
| 201 |
+
"learning_rate": 1.316348195329087e-06,
|
| 202 |
+
"loss": 0.9537,
|
| 203 |
+
"step": 310
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 2.0,
|
| 207 |
+
"eval_accuracy": 0.52,
|
| 208 |
+
"eval_loss": 0.9655842781066895,
|
| 209 |
+
"eval_runtime": 5.4121,
|
| 210 |
+
"eval_samples_per_second": 73.908,
|
| 211 |
+
"eval_steps_per_second": 9.239,
|
| 212 |
+
"step": 314
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 2.04,
|
| 216 |
+
"learning_rate": 1.3588110403397026e-06,
|
| 217 |
+
"loss": 0.9027,
|
| 218 |
+
"step": 320
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 2.1,
|
| 222 |
+
"learning_rate": 1.4012738853503183e-06,
|
| 223 |
+
"loss": 0.8836,
|
| 224 |
+
"step": 330
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 2.17,
|
| 228 |
+
"learning_rate": 1.4394904458598724e-06,
|
| 229 |
+
"loss": 0.8358,
|
| 230 |
+
"step": 340
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 2.23,
|
| 234 |
+
"learning_rate": 1.481953290870488e-06,
|
| 235 |
+
"loss": 0.8609,
|
| 236 |
+
"step": 350
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 2.29,
|
| 240 |
+
"learning_rate": 1.5244161358811042e-06,
|
| 241 |
+
"loss": 0.8037,
|
| 242 |
+
"step": 360
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 2.36,
|
| 246 |
+
"learning_rate": 1.5668789808917198e-06,
|
| 247 |
+
"loss": 0.7825,
|
| 248 |
+
"step": 370
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 2.42,
|
| 252 |
+
"learning_rate": 1.6093418259023355e-06,
|
| 253 |
+
"loss": 0.8254,
|
| 254 |
+
"step": 380
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 2.48,
|
| 258 |
+
"learning_rate": 1.6518046709129512e-06,
|
| 259 |
+
"loss": 0.79,
|
| 260 |
+
"step": 390
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 2.55,
|
| 264 |
+
"learning_rate": 1.6942675159235668e-06,
|
| 265 |
+
"loss": 0.8032,
|
| 266 |
+
"step": 400
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 2.61,
|
| 270 |
+
"learning_rate": 1.7367303609341825e-06,
|
| 271 |
+
"loss": 0.7757,
|
| 272 |
+
"step": 410
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.68,
|
| 276 |
+
"learning_rate": 1.7791932059447982e-06,
|
| 277 |
+
"loss": 0.7849,
|
| 278 |
+
"step": 420
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 2.74,
|
| 282 |
+
"learning_rate": 1.8216560509554139e-06,
|
| 283 |
+
"loss": 0.7449,
|
| 284 |
+
"step": 430
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 2.8,
|
| 288 |
+
"learning_rate": 1.8641188959660298e-06,
|
| 289 |
+
"loss": 0.7586,
|
| 290 |
+
"step": 440
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 2.87,
|
| 294 |
+
"learning_rate": 1.9065817409766454e-06,
|
| 295 |
+
"loss": 0.7547,
|
| 296 |
+
"step": 450
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 2.93,
|
| 300 |
+
"learning_rate": 1.949044585987261e-06,
|
| 301 |
+
"loss": 0.7729,
|
| 302 |
+
"step": 460
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 2.99,
|
| 306 |
+
"learning_rate": 1.987261146496815e-06,
|
| 307 |
+
"loss": 0.7124,
|
| 308 |
+
"step": 470
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 3.0,
|
| 312 |
+
"eval_accuracy": 0.575,
|
| 313 |
+
"eval_loss": 0.8280940055847168,
|
| 314 |
+
"eval_runtime": 5.3205,
|
| 315 |
+
"eval_samples_per_second": 75.18,
|
| 316 |
+
"eval_steps_per_second": 9.398,
|
| 317 |
+
"step": 471
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 3.06,
|
| 321 |
+
"learning_rate": 1.996697334276952e-06,
|
| 322 |
+
"loss": 0.7028,
|
| 323 |
+
"step": 480
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 3.12,
|
| 327 |
+
"learning_rate": 1.9919792403868834e-06,
|
| 328 |
+
"loss": 0.6914,
|
| 329 |
+
"step": 490
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 3.18,
|
| 333 |
+
"learning_rate": 1.987261146496815e-06,
|
| 334 |
+
"loss": 0.7107,
|
| 335 |
+
"step": 500
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 3.25,
|
| 339 |
+
"learning_rate": 1.9825430526067466e-06,
|
| 340 |
+
"loss": 0.7197,
|
| 341 |
+
"step": 510
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 3.31,
|
| 345 |
+
"learning_rate": 1.9778249587166784e-06,
|
| 346 |
+
"loss": 0.6729,
|
| 347 |
+
"step": 520
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 3.38,
|
| 351 |
+
"learning_rate": 1.97310686482661e-06,
|
| 352 |
+
"loss": 0.699,
|
| 353 |
+
"step": 530
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 3.44,
|
| 357 |
+
"learning_rate": 1.9683887709365417e-06,
|
| 358 |
+
"loss": 0.6529,
|
| 359 |
+
"step": 540
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 3.5,
|
| 363 |
+
"learning_rate": 1.963670677046473e-06,
|
| 364 |
+
"loss": 0.6093,
|
| 365 |
+
"step": 550
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 3.57,
|
| 369 |
+
"learning_rate": 1.9589525831564045e-06,
|
| 370 |
+
"loss": 0.6709,
|
| 371 |
+
"step": 560
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 3.63,
|
| 375 |
+
"learning_rate": 1.9542344892663363e-06,
|
| 376 |
+
"loss": 0.6327,
|
| 377 |
+
"step": 570
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 3.69,
|
| 381 |
+
"learning_rate": 1.9495163953762677e-06,
|
| 382 |
+
"loss": 0.6713,
|
| 383 |
+
"step": 580
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 3.76,
|
| 387 |
+
"learning_rate": 1.9447983014861995e-06,
|
| 388 |
+
"loss": 0.6506,
|
| 389 |
+
"step": 590
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 3.82,
|
| 393 |
+
"learning_rate": 1.940080207596131e-06,
|
| 394 |
+
"loss": 0.6314,
|
| 395 |
+
"step": 600
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 3.89,
|
| 399 |
+
"learning_rate": 1.9353621137060628e-06,
|
| 400 |
+
"loss": 0.6114,
|
| 401 |
+
"step": 610
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 3.95,
|
| 405 |
+
"learning_rate": 1.930644019815994e-06,
|
| 406 |
+
"loss": 0.6114,
|
| 407 |
+
"step": 620
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 4.0,
|
| 411 |
+
"eval_accuracy": 0.64,
|
| 412 |
+
"eval_loss": 0.753457248210907,
|
| 413 |
+
"eval_runtime": 5.5017,
|
| 414 |
+
"eval_samples_per_second": 72.704,
|
| 415 |
+
"eval_steps_per_second": 9.088,
|
| 416 |
+
"step": 628
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 4.01,
|
| 420 |
+
"learning_rate": 1.9259259259259256e-06,
|
| 421 |
+
"loss": 0.6592,
|
| 422 |
+
"step": 630
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 4.08,
|
| 426 |
+
"learning_rate": 1.9212078320358574e-06,
|
| 427 |
+
"loss": 0.5994,
|
| 428 |
+
"step": 640
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 4.14,
|
| 432 |
+
"learning_rate": 1.916489738145789e-06,
|
| 433 |
+
"loss": 0.6178,
|
| 434 |
+
"step": 650
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 4.2,
|
| 438 |
+
"learning_rate": 1.9117716442557206e-06,
|
| 439 |
+
"loss": 0.5675,
|
| 440 |
+
"step": 660
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 4.27,
|
| 444 |
+
"learning_rate": 1.9070535503656522e-06,
|
| 445 |
+
"loss": 0.6167,
|
| 446 |
+
"step": 670
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 4.33,
|
| 450 |
+
"learning_rate": 1.9023354564755839e-06,
|
| 451 |
+
"loss": 0.5665,
|
| 452 |
+
"step": 680
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 4.39,
|
| 456 |
+
"learning_rate": 1.8976173625855155e-06,
|
| 457 |
+
"loss": 0.6061,
|
| 458 |
+
"step": 690
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 4.46,
|
| 462 |
+
"learning_rate": 1.892899268695447e-06,
|
| 463 |
+
"loss": 0.5977,
|
| 464 |
+
"step": 700
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 4.52,
|
| 468 |
+
"learning_rate": 1.8881811748053787e-06,
|
| 469 |
+
"loss": 0.5113,
|
| 470 |
+
"step": 710
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 4.59,
|
| 474 |
+
"learning_rate": 1.8834630809153101e-06,
|
| 475 |
+
"loss": 0.5353,
|
| 476 |
+
"step": 720
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 4.65,
|
| 480 |
+
"learning_rate": 1.8787449870252417e-06,
|
| 481 |
+
"loss": 0.5925,
|
| 482 |
+
"step": 730
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 4.71,
|
| 486 |
+
"learning_rate": 1.8740268931351733e-06,
|
| 487 |
+
"loss": 0.5639,
|
| 488 |
+
"step": 740
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 4.78,
|
| 492 |
+
"learning_rate": 1.869308799245105e-06,
|
| 493 |
+
"loss": 0.5562,
|
| 494 |
+
"step": 750
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 4.84,
|
| 498 |
+
"learning_rate": 1.8645907053550366e-06,
|
| 499 |
+
"loss": 0.544,
|
| 500 |
+
"step": 760
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 4.9,
|
| 504 |
+
"learning_rate": 1.8598726114649682e-06,
|
| 505 |
+
"loss": 0.5563,
|
| 506 |
+
"step": 770
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 4.97,
|
| 510 |
+
"learning_rate": 1.8551545175748998e-06,
|
| 511 |
+
"loss": 0.517,
|
| 512 |
+
"step": 780
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 5.0,
|
| 516 |
+
"eval_accuracy": 0.725,
|
| 517 |
+
"eval_loss": 0.6396925449371338,
|
| 518 |
+
"eval_runtime": 5.3831,
|
| 519 |
+
"eval_samples_per_second": 74.306,
|
| 520 |
+
"eval_steps_per_second": 9.288,
|
| 521 |
+
"step": 785
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 5.03,
|
| 525 |
+
"learning_rate": 1.8504364236848314e-06,
|
| 526 |
+
"loss": 0.5165,
|
| 527 |
+
"step": 790
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 5.1,
|
| 531 |
+
"learning_rate": 1.845718329794763e-06,
|
| 532 |
+
"loss": 0.5175,
|
| 533 |
+
"step": 800
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 5.16,
|
| 537 |
+
"learning_rate": 1.8410002359046944e-06,
|
| 538 |
+
"loss": 0.5361,
|
| 539 |
+
"step": 810
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 5.22,
|
| 543 |
+
"learning_rate": 1.8362821420146258e-06,
|
| 544 |
+
"loss": 0.475,
|
| 545 |
+
"step": 820
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 5.29,
|
| 549 |
+
"learning_rate": 1.8315640481245575e-06,
|
| 550 |
+
"loss": 0.4947,
|
| 551 |
+
"step": 830
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 5.35,
|
| 555 |
+
"learning_rate": 1.826845954234489e-06,
|
| 556 |
+
"loss": 0.504,
|
| 557 |
+
"step": 840
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 5.41,
|
| 561 |
+
"learning_rate": 1.8221278603444207e-06,
|
| 562 |
+
"loss": 0.5223,
|
| 563 |
+
"step": 850
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 5.48,
|
| 567 |
+
"learning_rate": 1.8174097664543523e-06,
|
| 568 |
+
"loss": 0.4913,
|
| 569 |
+
"step": 860
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 5.54,
|
| 573 |
+
"learning_rate": 1.812691672564284e-06,
|
| 574 |
+
"loss": 0.4583,
|
| 575 |
+
"step": 870
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 5.61,
|
| 579 |
+
"learning_rate": 1.8079735786742155e-06,
|
| 580 |
+
"loss": 0.4675,
|
| 581 |
+
"step": 880
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 5.67,
|
| 585 |
+
"learning_rate": 1.803255484784147e-06,
|
| 586 |
+
"loss": 0.4712,
|
| 587 |
+
"step": 890
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 5.73,
|
| 591 |
+
"learning_rate": 1.7985373908940786e-06,
|
| 592 |
+
"loss": 0.5068,
|
| 593 |
+
"step": 900
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 5.8,
|
| 597 |
+
"learning_rate": 1.7938192970040102e-06,
|
| 598 |
+
"loss": 0.5049,
|
| 599 |
+
"step": 910
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 5.86,
|
| 603 |
+
"learning_rate": 1.7891012031139418e-06,
|
| 604 |
+
"loss": 0.4855,
|
| 605 |
+
"step": 920
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 5.92,
|
| 609 |
+
"learning_rate": 1.7843831092238734e-06,
|
| 610 |
+
"loss": 0.4233,
|
| 611 |
+
"step": 930
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 5.99,
|
| 615 |
+
"learning_rate": 1.779665015333805e-06,
|
| 616 |
+
"loss": 0.4773,
|
| 617 |
+
"step": 940
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 6.0,
|
| 621 |
+
"eval_accuracy": 0.78,
|
| 622 |
+
"eval_loss": 0.5554624795913696,
|
| 623 |
+
"eval_runtime": 5.952,
|
| 624 |
+
"eval_samples_per_second": 67.204,
|
| 625 |
+
"eval_steps_per_second": 8.4,
|
| 626 |
+
"step": 942
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 6.05,
|
| 630 |
+
"learning_rate": 1.7749469214437366e-06,
|
| 631 |
+
"loss": 0.455,
|
| 632 |
+
"step": 950
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 6.11,
|
| 636 |
+
"learning_rate": 1.7702288275536682e-06,
|
| 637 |
+
"loss": 0.4656,
|
| 638 |
+
"step": 960
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 6.18,
|
| 642 |
+
"learning_rate": 1.7655107336635999e-06,
|
| 643 |
+
"loss": 0.4254,
|
| 644 |
+
"step": 970
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 6.24,
|
| 648 |
+
"learning_rate": 1.7607926397735313e-06,
|
| 649 |
+
"loss": 0.4177,
|
| 650 |
+
"step": 980
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 6.31,
|
| 654 |
+
"learning_rate": 1.7560745458834629e-06,
|
| 655 |
+
"loss": 0.4008,
|
| 656 |
+
"step": 990
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 6.37,
|
| 660 |
+
"learning_rate": 1.7513564519933945e-06,
|
| 661 |
+
"loss": 0.449,
|
| 662 |
+
"step": 1000
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 6.43,
|
| 666 |
+
"learning_rate": 1.7466383581033261e-06,
|
| 667 |
+
"loss": 0.4398,
|
| 668 |
+
"step": 1010
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 6.5,
|
| 672 |
+
"learning_rate": 1.7419202642132577e-06,
|
| 673 |
+
"loss": 0.4012,
|
| 674 |
+
"step": 1020
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 6.56,
|
| 678 |
+
"learning_rate": 1.7372021703231893e-06,
|
| 679 |
+
"loss": 0.3977,
|
| 680 |
+
"step": 1030
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 6.62,
|
| 684 |
+
"learning_rate": 1.732484076433121e-06,
|
| 685 |
+
"loss": 0.4268,
|
| 686 |
+
"step": 1040
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 6.69,
|
| 690 |
+
"learning_rate": 1.7277659825430526e-06,
|
| 691 |
+
"loss": 0.4132,
|
| 692 |
+
"step": 1050
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 6.75,
|
| 696 |
+
"learning_rate": 1.7230478886529842e-06,
|
| 697 |
+
"loss": 0.3898,
|
| 698 |
+
"step": 1060
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 6.82,
|
| 702 |
+
"learning_rate": 1.7183297947629156e-06,
|
| 703 |
+
"loss": 0.4145,
|
| 704 |
+
"step": 1070
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 6.88,
|
| 708 |
+
"learning_rate": 1.7136117008728472e-06,
|
| 709 |
+
"loss": 0.3537,
|
| 710 |
+
"step": 1080
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 6.94,
|
| 714 |
+
"learning_rate": 1.7088936069827788e-06,
|
| 715 |
+
"loss": 0.402,
|
| 716 |
+
"step": 1090
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 7.0,
|
| 720 |
+
"eval_accuracy": 0.845,
|
| 721 |
+
"eval_loss": 0.4491625130176544,
|
| 722 |
+
"eval_runtime": 5.5591,
|
| 723 |
+
"eval_samples_per_second": 71.954,
|
| 724 |
+
"eval_steps_per_second": 8.994,
|
| 725 |
+
"step": 1099
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 7.01,
|
| 729 |
+
"learning_rate": 1.7041755130927104e-06,
|
| 730 |
+
"loss": 0.376,
|
| 731 |
+
"step": 1100
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 7.07,
|
| 735 |
+
"learning_rate": 1.699457419202642e-06,
|
| 736 |
+
"loss": 0.3519,
|
| 737 |
+
"step": 1110
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 7.13,
|
| 741 |
+
"learning_rate": 1.6947393253125737e-06,
|
| 742 |
+
"loss": 0.3645,
|
| 743 |
+
"step": 1120
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 7.2,
|
| 747 |
+
"learning_rate": 1.6900212314225053e-06,
|
| 748 |
+
"loss": 0.3703,
|
| 749 |
+
"step": 1130
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 7.26,
|
| 753 |
+
"learning_rate": 1.685303137532437e-06,
|
| 754 |
+
"loss": 0.3766,
|
| 755 |
+
"step": 1140
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 7.32,
|
| 759 |
+
"learning_rate": 1.6805850436423683e-06,
|
| 760 |
+
"loss": 0.3608,
|
| 761 |
+
"step": 1150
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 7.39,
|
| 765 |
+
"learning_rate": 1.6758669497523e-06,
|
| 766 |
+
"loss": 0.3515,
|
| 767 |
+
"step": 1160
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 7.45,
|
| 771 |
+
"learning_rate": 1.6711488558622315e-06,
|
| 772 |
+
"loss": 0.3624,
|
| 773 |
+
"step": 1170
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 7.52,
|
| 777 |
+
"learning_rate": 1.6664307619721631e-06,
|
| 778 |
+
"loss": 0.3434,
|
| 779 |
+
"step": 1180
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 7.58,
|
| 783 |
+
"learning_rate": 1.6617126680820948e-06,
|
| 784 |
+
"loss": 0.3247,
|
| 785 |
+
"step": 1190
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 7.64,
|
| 789 |
+
"learning_rate": 1.6569945741920264e-06,
|
| 790 |
+
"loss": 0.3103,
|
| 791 |
+
"step": 1200
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 7.71,
|
| 795 |
+
"learning_rate": 1.652276480301958e-06,
|
| 796 |
+
"loss": 0.3281,
|
| 797 |
+
"step": 1210
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 7.77,
|
| 801 |
+
"learning_rate": 1.6475583864118896e-06,
|
| 802 |
+
"loss": 0.3692,
|
| 803 |
+
"step": 1220
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 7.83,
|
| 807 |
+
"learning_rate": 1.6428402925218212e-06,
|
| 808 |
+
"loss": 0.3082,
|
| 809 |
+
"step": 1230
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 7.9,
|
| 813 |
+
"learning_rate": 1.6381221986317526e-06,
|
| 814 |
+
"loss": 0.3278,
|
| 815 |
+
"step": 1240
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 7.96,
|
| 819 |
+
"learning_rate": 1.6334041047416842e-06,
|
| 820 |
+
"loss": 0.3504,
|
| 821 |
+
"step": 1250
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 8.0,
|
| 825 |
+
"eval_accuracy": 0.8775,
|
| 826 |
+
"eval_loss": 0.37387552857398987,
|
| 827 |
+
"eval_runtime": 6.4049,
|
| 828 |
+
"eval_samples_per_second": 62.452,
|
| 829 |
+
"eval_steps_per_second": 7.807,
|
| 830 |
+
"step": 1256
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 8.03,
|
| 834 |
+
"learning_rate": 1.6286860108516159e-06,
|
| 835 |
+
"loss": 0.3603,
|
| 836 |
+
"step": 1260
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 8.09,
|
| 840 |
+
"learning_rate": 1.6239679169615475e-06,
|
| 841 |
+
"loss": 0.2868,
|
| 842 |
+
"step": 1270
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 8.15,
|
| 846 |
+
"learning_rate": 1.619249823071479e-06,
|
| 847 |
+
"loss": 0.292,
|
| 848 |
+
"step": 1280
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 8.22,
|
| 852 |
+
"learning_rate": 1.6145317291814107e-06,
|
| 853 |
+
"loss": 0.321,
|
| 854 |
+
"step": 1290
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 8.28,
|
| 858 |
+
"learning_rate": 1.6098136352913423e-06,
|
| 859 |
+
"loss": 0.3054,
|
| 860 |
+
"step": 1300
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 8.34,
|
| 864 |
+
"learning_rate": 1.605095541401274e-06,
|
| 865 |
+
"loss": 0.2833,
|
| 866 |
+
"step": 1310
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 8.41,
|
| 870 |
+
"learning_rate": 1.6003774475112053e-06,
|
| 871 |
+
"loss": 0.3144,
|
| 872 |
+
"step": 1320
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 8.47,
|
| 876 |
+
"learning_rate": 1.595659353621137e-06,
|
| 877 |
+
"loss": 0.2801,
|
| 878 |
+
"step": 1330
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 8.54,
|
| 882 |
+
"learning_rate": 1.5909412597310686e-06,
|
| 883 |
+
"loss": 0.2938,
|
| 884 |
+
"step": 1340
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 8.6,
|
| 888 |
+
"learning_rate": 1.5862231658410002e-06,
|
| 889 |
+
"loss": 0.2794,
|
| 890 |
+
"step": 1350
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 8.66,
|
| 894 |
+
"learning_rate": 1.5815050719509318e-06,
|
| 895 |
+
"loss": 0.2566,
|
| 896 |
+
"step": 1360
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 8.73,
|
| 900 |
+
"learning_rate": 1.5767869780608634e-06,
|
| 901 |
+
"loss": 0.2926,
|
| 902 |
+
"step": 1370
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 8.79,
|
| 906 |
+
"learning_rate": 1.572068884170795e-06,
|
| 907 |
+
"loss": 0.2733,
|
| 908 |
+
"step": 1380
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 8.85,
|
| 912 |
+
"learning_rate": 1.5673507902807266e-06,
|
| 913 |
+
"loss": 0.2815,
|
| 914 |
+
"step": 1390
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 8.92,
|
| 918 |
+
"learning_rate": 1.5626326963906583e-06,
|
| 919 |
+
"loss": 0.2973,
|
| 920 |
+
"step": 1400
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 8.98,
|
| 924 |
+
"learning_rate": 1.5579146025005897e-06,
|
| 925 |
+
"loss": 0.2584,
|
| 926 |
+
"step": 1410
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 9.0,
|
| 930 |
+
"eval_accuracy": 0.89,
|
| 931 |
+
"eval_loss": 0.3347581624984741,
|
| 932 |
+
"eval_runtime": 6.0288,
|
| 933 |
+
"eval_samples_per_second": 66.348,
|
| 934 |
+
"eval_steps_per_second": 8.294,
|
| 935 |
+
"step": 1413
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 9.04,
|
| 939 |
+
"learning_rate": 1.5531965086105213e-06,
|
| 940 |
+
"loss": 0.2353,
|
| 941 |
+
"step": 1420
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 9.11,
|
| 945 |
+
"learning_rate": 1.548478414720453e-06,
|
| 946 |
+
"loss": 0.2659,
|
| 947 |
+
"step": 1430
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 9.17,
|
| 951 |
+
"learning_rate": 1.5437603208303845e-06,
|
| 952 |
+
"loss": 0.2395,
|
| 953 |
+
"step": 1440
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 9.24,
|
| 957 |
+
"learning_rate": 1.5390422269403161e-06,
|
| 958 |
+
"loss": 0.2494,
|
| 959 |
+
"step": 1450
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 9.3,
|
| 963 |
+
"learning_rate": 1.5343241330502477e-06,
|
| 964 |
+
"loss": 0.2551,
|
| 965 |
+
"step": 1460
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 9.36,
|
| 969 |
+
"learning_rate": 1.5296060391601794e-06,
|
| 970 |
+
"loss": 0.2346,
|
| 971 |
+
"step": 1470
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 9.43,
|
| 975 |
+
"learning_rate": 1.524887945270111e-06,
|
| 976 |
+
"loss": 0.2311,
|
| 977 |
+
"step": 1480
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 9.49,
|
| 981 |
+
"learning_rate": 1.5201698513800426e-06,
|
| 982 |
+
"loss": 0.237,
|
| 983 |
+
"step": 1490
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 9.55,
|
| 987 |
+
"learning_rate": 1.515451757489974e-06,
|
| 988 |
+
"loss": 0.2366,
|
| 989 |
+
"step": 1500
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 9.62,
|
| 993 |
+
"learning_rate": 1.5107336635999056e-06,
|
| 994 |
+
"loss": 0.242,
|
| 995 |
+
"step": 1510
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 9.68,
|
| 999 |
+
"learning_rate": 1.5060155697098372e-06,
|
| 1000 |
+
"loss": 0.2635,
|
| 1001 |
+
"step": 1520
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 9.75,
|
| 1005 |
+
"learning_rate": 1.5012974758197688e-06,
|
| 1006 |
+
"loss": 0.2243,
|
| 1007 |
+
"step": 1530
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 9.81,
|
| 1011 |
+
"learning_rate": 1.4965793819297002e-06,
|
| 1012 |
+
"loss": 0.236,
|
| 1013 |
+
"step": 1540
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 9.87,
|
| 1017 |
+
"learning_rate": 1.4918612880396319e-06,
|
| 1018 |
+
"loss": 0.2312,
|
| 1019 |
+
"step": 1550
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 9.94,
|
| 1023 |
+
"learning_rate": 1.4871431941495635e-06,
|
| 1024 |
+
"loss": 0.2357,
|
| 1025 |
+
"step": 1560
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 10.0,
|
| 1029 |
+
"learning_rate": 1.482425100259495e-06,
|
| 1030 |
+
"loss": 0.2439,
|
| 1031 |
+
"step": 1570
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 10.0,
|
| 1035 |
+
"eval_accuracy": 0.9,
|
| 1036 |
+
"eval_loss": 0.29536351561546326,
|
| 1037 |
+
"eval_runtime": 6.1912,
|
| 1038 |
+
"eval_samples_per_second": 64.608,
|
| 1039 |
+
"eval_steps_per_second": 8.076,
|
| 1040 |
+
"step": 1570
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 10.06,
|
| 1044 |
+
"learning_rate": 1.4777070063694265e-06,
|
| 1045 |
+
"loss": 0.2089,
|
| 1046 |
+
"step": 1580
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 10.13,
|
| 1050 |
+
"learning_rate": 1.4729889124793581e-06,
|
| 1051 |
+
"loss": 0.198,
|
| 1052 |
+
"step": 1590
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 10.19,
|
| 1056 |
+
"learning_rate": 1.4682708185892897e-06,
|
| 1057 |
+
"loss": 0.2017,
|
| 1058 |
+
"step": 1600
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 10.25,
|
| 1062 |
+
"learning_rate": 1.4635527246992213e-06,
|
| 1063 |
+
"loss": 0.2091,
|
| 1064 |
+
"step": 1610
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 10.32,
|
| 1068 |
+
"learning_rate": 1.458834630809153e-06,
|
| 1069 |
+
"loss": 0.2016,
|
| 1070 |
+
"step": 1620
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 10.38,
|
| 1074 |
+
"learning_rate": 1.4541165369190846e-06,
|
| 1075 |
+
"loss": 0.2036,
|
| 1076 |
+
"step": 1630
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 10.45,
|
| 1080 |
+
"learning_rate": 1.4493984430290162e-06,
|
| 1081 |
+
"loss": 0.2159,
|
| 1082 |
+
"step": 1640
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 10.51,
|
| 1086 |
+
"learning_rate": 1.4446803491389478e-06,
|
| 1087 |
+
"loss": 0.2042,
|
| 1088 |
+
"step": 1650
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 10.57,
|
| 1092 |
+
"learning_rate": 1.4399622552488794e-06,
|
| 1093 |
+
"loss": 0.1891,
|
| 1094 |
+
"step": 1660
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 10.64,
|
| 1098 |
+
"learning_rate": 1.4352441613588108e-06,
|
| 1099 |
+
"loss": 0.196,
|
| 1100 |
+
"step": 1670
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 10.7,
|
| 1104 |
+
"learning_rate": 1.4305260674687424e-06,
|
| 1105 |
+
"loss": 0.1919,
|
| 1106 |
+
"step": 1680
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 10.76,
|
| 1110 |
+
"learning_rate": 1.425807973578674e-06,
|
| 1111 |
+
"loss": 0.1908,
|
| 1112 |
+
"step": 1690
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 10.83,
|
| 1116 |
+
"learning_rate": 1.4210898796886057e-06,
|
| 1117 |
+
"loss": 0.2065,
|
| 1118 |
+
"step": 1700
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 10.89,
|
| 1122 |
+
"learning_rate": 1.4163717857985373e-06,
|
| 1123 |
+
"loss": 0.1718,
|
| 1124 |
+
"step": 1710
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 10.96,
|
| 1128 |
+
"learning_rate": 1.411653691908469e-06,
|
| 1129 |
+
"loss": 0.1925,
|
| 1130 |
+
"step": 1720
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 11.0,
|
| 1134 |
+
"eval_accuracy": 0.9275,
|
| 1135 |
+
"eval_loss": 0.24589473009109497,
|
| 1136 |
+
"eval_runtime": 9.3749,
|
| 1137 |
+
"eval_samples_per_second": 42.667,
|
| 1138 |
+
"eval_steps_per_second": 5.333,
|
| 1139 |
+
"step": 1727
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 11.02,
|
| 1143 |
+
"learning_rate": 1.4069355980184005e-06,
|
| 1144 |
+
"loss": 0.2006,
|
| 1145 |
+
"step": 1730
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 11.08,
|
| 1149 |
+
"learning_rate": 1.4022175041283321e-06,
|
| 1150 |
+
"loss": 0.1676,
|
| 1151 |
+
"step": 1740
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 11.15,
|
| 1155 |
+
"learning_rate": 1.3974994102382637e-06,
|
| 1156 |
+
"loss": 0.1749,
|
| 1157 |
+
"step": 1750
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 11.21,
|
| 1161 |
+
"learning_rate": 1.3927813163481951e-06,
|
| 1162 |
+
"loss": 0.1616,
|
| 1163 |
+
"step": 1760
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 11.27,
|
| 1167 |
+
"learning_rate": 1.3880632224581268e-06,
|
| 1168 |
+
"loss": 0.1564,
|
| 1169 |
+
"step": 1770
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 11.34,
|
| 1173 |
+
"learning_rate": 1.3833451285680584e-06,
|
| 1174 |
+
"loss": 0.1709,
|
| 1175 |
+
"step": 1780
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 11.4,
|
| 1179 |
+
"learning_rate": 1.37862703467799e-06,
|
| 1180 |
+
"loss": 0.1612,
|
| 1181 |
+
"step": 1790
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 11.46,
|
| 1185 |
+
"learning_rate": 1.3739089407879216e-06,
|
| 1186 |
+
"loss": 0.1634,
|
| 1187 |
+
"step": 1800
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 11.53,
|
| 1191 |
+
"learning_rate": 1.3691908468978532e-06,
|
| 1192 |
+
"loss": 0.1604,
|
| 1193 |
+
"step": 1810
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 11.59,
|
| 1197 |
+
"learning_rate": 1.3644727530077848e-06,
|
| 1198 |
+
"loss": 0.167,
|
| 1199 |
+
"step": 1820
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 11.66,
|
| 1203 |
+
"learning_rate": 1.3597546591177165e-06,
|
| 1204 |
+
"loss": 0.1548,
|
| 1205 |
+
"step": 1830
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 11.72,
|
| 1209 |
+
"learning_rate": 1.3550365652276479e-06,
|
| 1210 |
+
"loss": 0.1718,
|
| 1211 |
+
"step": 1840
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 11.78,
|
| 1215 |
+
"learning_rate": 1.3503184713375795e-06,
|
| 1216 |
+
"loss": 0.1443,
|
| 1217 |
+
"step": 1850
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 11.85,
|
| 1221 |
+
"learning_rate": 1.345600377447511e-06,
|
| 1222 |
+
"loss": 0.1606,
|
| 1223 |
+
"step": 1860
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 11.91,
|
| 1227 |
+
"learning_rate": 1.3408822835574427e-06,
|
| 1228 |
+
"loss": 0.1703,
|
| 1229 |
+
"step": 1870
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 11.97,
|
| 1233 |
+
"learning_rate": 1.3361641896673743e-06,
|
| 1234 |
+
"loss": 0.1693,
|
| 1235 |
+
"step": 1880
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 12.0,
|
| 1239 |
+
"eval_accuracy": 0.9325,
|
| 1240 |
+
"eval_loss": 0.21917079389095306,
|
| 1241 |
+
"eval_runtime": 5.5809,
|
| 1242 |
+
"eval_samples_per_second": 71.673,
|
| 1243 |
+
"eval_steps_per_second": 8.959,
|
| 1244 |
+
"step": 1884
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 12.04,
|
| 1248 |
+
"learning_rate": 1.331446095777306e-06,
|
| 1249 |
+
"loss": 0.1501,
|
| 1250 |
+
"step": 1890
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 12.1,
|
| 1254 |
+
"learning_rate": 1.3267280018872375e-06,
|
| 1255 |
+
"loss": 0.1339,
|
| 1256 |
+
"step": 1900
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 12.17,
|
| 1260 |
+
"learning_rate": 1.3220099079971692e-06,
|
| 1261 |
+
"loss": 0.1506,
|
| 1262 |
+
"step": 1910
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 12.23,
|
| 1266 |
+
"learning_rate": 1.3172918141071008e-06,
|
| 1267 |
+
"loss": 0.1347,
|
| 1268 |
+
"step": 1920
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 12.29,
|
| 1272 |
+
"learning_rate": 1.3125737202170322e-06,
|
| 1273 |
+
"loss": 0.1378,
|
| 1274 |
+
"step": 1930
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 12.36,
|
| 1278 |
+
"learning_rate": 1.3078556263269638e-06,
|
| 1279 |
+
"loss": 0.1367,
|
| 1280 |
+
"step": 1940
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 12.42,
|
| 1284 |
+
"learning_rate": 1.3031375324368954e-06,
|
| 1285 |
+
"loss": 0.1236,
|
| 1286 |
+
"step": 1950
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 12.48,
|
| 1290 |
+
"learning_rate": 1.298419438546827e-06,
|
| 1291 |
+
"loss": 0.1332,
|
| 1292 |
+
"step": 1960
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 12.55,
|
| 1296 |
+
"learning_rate": 1.2937013446567586e-06,
|
| 1297 |
+
"loss": 0.1424,
|
| 1298 |
+
"step": 1970
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 12.61,
|
| 1302 |
+
"learning_rate": 1.2889832507666903e-06,
|
| 1303 |
+
"loss": 0.1426,
|
| 1304 |
+
"step": 1980
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 12.68,
|
| 1308 |
+
"learning_rate": 1.2842651568766219e-06,
|
| 1309 |
+
"loss": 0.1364,
|
| 1310 |
+
"step": 1990
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 12.74,
|
| 1314 |
+
"learning_rate": 1.2795470629865535e-06,
|
| 1315 |
+
"loss": 0.1488,
|
| 1316 |
+
"step": 2000
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 12.8,
|
| 1320 |
+
"learning_rate": 1.274828969096485e-06,
|
| 1321 |
+
"loss": 0.1308,
|
| 1322 |
+
"step": 2010
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 12.87,
|
| 1326 |
+
"learning_rate": 1.2701108752064165e-06,
|
| 1327 |
+
"loss": 0.1278,
|
| 1328 |
+
"step": 2020
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 12.93,
|
| 1332 |
+
"learning_rate": 1.2653927813163481e-06,
|
| 1333 |
+
"loss": 0.1134,
|
| 1334 |
+
"step": 2030
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 12.99,
|
| 1338 |
+
"learning_rate": 1.2606746874262797e-06,
|
| 1339 |
+
"loss": 0.1157,
|
| 1340 |
+
"step": 2040
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 13.0,
|
| 1344 |
+
"eval_accuracy": 0.9375,
|
| 1345 |
+
"eval_loss": 0.22450672090053558,
|
| 1346 |
+
"eval_runtime": 5.5269,
|
| 1347 |
+
"eval_samples_per_second": 72.373,
|
| 1348 |
+
"eval_steps_per_second": 9.047,
|
| 1349 |
+
"step": 2041
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 13.06,
|
| 1353 |
+
"learning_rate": 1.2559565935362114e-06,
|
| 1354 |
+
"loss": 0.1235,
|
| 1355 |
+
"step": 2050
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 13.12,
|
| 1359 |
+
"learning_rate": 1.251238499646143e-06,
|
| 1360 |
+
"loss": 0.117,
|
| 1361 |
+
"step": 2060
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 13.18,
|
| 1365 |
+
"learning_rate": 1.2465204057560746e-06,
|
| 1366 |
+
"loss": 0.1192,
|
| 1367 |
+
"step": 2070
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 13.25,
|
| 1371 |
+
"learning_rate": 1.2418023118660062e-06,
|
| 1372 |
+
"loss": 0.1031,
|
| 1373 |
+
"step": 2080
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 13.31,
|
| 1377 |
+
"learning_rate": 1.2370842179759378e-06,
|
| 1378 |
+
"loss": 0.1111,
|
| 1379 |
+
"step": 2090
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 13.38,
|
| 1383 |
+
"learning_rate": 1.2323661240858692e-06,
|
| 1384 |
+
"loss": 0.0991,
|
| 1385 |
+
"step": 2100
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 13.44,
|
| 1389 |
+
"learning_rate": 1.2276480301958008e-06,
|
| 1390 |
+
"loss": 0.1041,
|
| 1391 |
+
"step": 2110
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 13.5,
|
| 1395 |
+
"learning_rate": 1.2229299363057325e-06,
|
| 1396 |
+
"loss": 0.1251,
|
| 1397 |
+
"step": 2120
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 13.57,
|
| 1401 |
+
"learning_rate": 1.218211842415664e-06,
|
| 1402 |
+
"loss": 0.0937,
|
| 1403 |
+
"step": 2130
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 13.63,
|
| 1407 |
+
"learning_rate": 1.2134937485255957e-06,
|
| 1408 |
+
"loss": 0.1133,
|
| 1409 |
+
"step": 2140
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 13.69,
|
| 1413 |
+
"learning_rate": 1.2087756546355273e-06,
|
| 1414 |
+
"loss": 0.0928,
|
| 1415 |
+
"step": 2150
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 13.76,
|
| 1419 |
+
"learning_rate": 1.204057560745459e-06,
|
| 1420 |
+
"loss": 0.1286,
|
| 1421 |
+
"step": 2160
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 13.82,
|
| 1425 |
+
"learning_rate": 1.1993394668553905e-06,
|
| 1426 |
+
"loss": 0.102,
|
| 1427 |
+
"step": 2170
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 13.89,
|
| 1431 |
+
"learning_rate": 1.1946213729653221e-06,
|
| 1432 |
+
"loss": 0.1028,
|
| 1433 |
+
"step": 2180
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 13.95,
|
| 1437 |
+
"learning_rate": 1.1899032790752535e-06,
|
| 1438 |
+
"loss": 0.1014,
|
| 1439 |
+
"step": 2190
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 14.0,
|
| 1443 |
+
"eval_accuracy": 0.9475,
|
| 1444 |
+
"eval_loss": 0.18889574706554413,
|
| 1445 |
+
"eval_runtime": 5.4539,
|
| 1446 |
+
"eval_samples_per_second": 73.342,
|
| 1447 |
+
"eval_steps_per_second": 9.168,
|
| 1448 |
+
"step": 2198
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 14.01,
|
| 1452 |
+
"learning_rate": 1.1851851851851852e-06,
|
| 1453 |
+
"loss": 0.112,
|
| 1454 |
+
"step": 2200
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 14.08,
|
| 1458 |
+
"learning_rate": 1.1804670912951168e-06,
|
| 1459 |
+
"loss": 0.0879,
|
| 1460 |
+
"step": 2210
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 14.14,
|
| 1464 |
+
"learning_rate": 1.1757489974050484e-06,
|
| 1465 |
+
"loss": 0.089,
|
| 1466 |
+
"step": 2220
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 14.2,
|
| 1470 |
+
"learning_rate": 1.17103090351498e-06,
|
| 1471 |
+
"loss": 0.0844,
|
| 1472 |
+
"step": 2230
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 14.27,
|
| 1476 |
+
"learning_rate": 1.1663128096249116e-06,
|
| 1477 |
+
"loss": 0.0949,
|
| 1478 |
+
"step": 2240
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 14.33,
|
| 1482 |
+
"learning_rate": 1.161594715734843e-06,
|
| 1483 |
+
"loss": 0.1087,
|
| 1484 |
+
"step": 2250
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 14.39,
|
| 1488 |
+
"learning_rate": 1.1568766218447746e-06,
|
| 1489 |
+
"loss": 0.0824,
|
| 1490 |
+
"step": 2260
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 14.46,
|
| 1494 |
+
"learning_rate": 1.152158527954706e-06,
|
| 1495 |
+
"loss": 0.0885,
|
| 1496 |
+
"step": 2270
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 14.52,
|
| 1500 |
+
"learning_rate": 1.1474404340646377e-06,
|
| 1501 |
+
"loss": 0.0886,
|
| 1502 |
+
"step": 2280
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 14.59,
|
| 1506 |
+
"learning_rate": 1.1427223401745693e-06,
|
| 1507 |
+
"loss": 0.1076,
|
| 1508 |
+
"step": 2290
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 14.65,
|
| 1512 |
+
"learning_rate": 1.1380042462845009e-06,
|
| 1513 |
+
"loss": 0.0756,
|
| 1514 |
+
"step": 2300
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 14.71,
|
| 1518 |
+
"learning_rate": 1.1332861523944325e-06,
|
| 1519 |
+
"loss": 0.0855,
|
| 1520 |
+
"step": 2310
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 14.78,
|
| 1524 |
+
"learning_rate": 1.1285680585043641e-06,
|
| 1525 |
+
"loss": 0.0811,
|
| 1526 |
+
"step": 2320
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 14.84,
|
| 1530 |
+
"learning_rate": 1.1238499646142957e-06,
|
| 1531 |
+
"loss": 0.0884,
|
| 1532 |
+
"step": 2330
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 14.9,
|
| 1536 |
+
"learning_rate": 1.1191318707242274e-06,
|
| 1537 |
+
"loss": 0.0899,
|
| 1538 |
+
"step": 2340
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 14.97,
|
| 1542 |
+
"learning_rate": 1.114413776834159e-06,
|
| 1543 |
+
"loss": 0.075,
|
| 1544 |
+
"step": 2350
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 15.0,
|
| 1548 |
+
"eval_accuracy": 0.9525,
|
| 1549 |
+
"eval_loss": 0.17460297048091888,
|
| 1550 |
+
"eval_runtime": 5.5354,
|
| 1551 |
+
"eval_samples_per_second": 72.262,
|
| 1552 |
+
"eval_steps_per_second": 9.033,
|
| 1553 |
+
"step": 2355
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 15.03,
|
| 1557 |
+
"learning_rate": 1.1096956829440904e-06,
|
| 1558 |
+
"loss": 0.0766,
|
| 1559 |
+
"step": 2360
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 15.1,
|
| 1563 |
+
"learning_rate": 1.104977589054022e-06,
|
| 1564 |
+
"loss": 0.0702,
|
| 1565 |
+
"step": 2370
|
| 1566 |
+
},
|
| 1567 |
+
{
|
| 1568 |
+
"epoch": 15.16,
|
| 1569 |
+
"learning_rate": 1.1002594951639536e-06,
|
| 1570 |
+
"loss": 0.0888,
|
| 1571 |
+
"step": 2380
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 15.22,
|
| 1575 |
+
"learning_rate": 1.0955414012738852e-06,
|
| 1576 |
+
"loss": 0.0716,
|
| 1577 |
+
"step": 2390
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 15.29,
|
| 1581 |
+
"learning_rate": 1.0908233073838168e-06,
|
| 1582 |
+
"loss": 0.0687,
|
| 1583 |
+
"step": 2400
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 15.35,
|
| 1587 |
+
"learning_rate": 1.0861052134937485e-06,
|
| 1588 |
+
"loss": 0.073,
|
| 1589 |
+
"step": 2410
|
| 1590 |
+
},
|
| 1591 |
+
{
|
| 1592 |
+
"epoch": 15.41,
|
| 1593 |
+
"learning_rate": 1.08138711960368e-06,
|
| 1594 |
+
"loss": 0.0695,
|
| 1595 |
+
"step": 2420
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 15.48,
|
| 1599 |
+
"learning_rate": 1.0766690257136117e-06,
|
| 1600 |
+
"loss": 0.0738,
|
| 1601 |
+
"step": 2430
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 15.54,
|
| 1605 |
+
"learning_rate": 1.0719509318235433e-06,
|
| 1606 |
+
"loss": 0.0836,
|
| 1607 |
+
"step": 2440
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 15.61,
|
| 1611 |
+
"learning_rate": 1.0672328379334747e-06,
|
| 1612 |
+
"loss": 0.0622,
|
| 1613 |
+
"step": 2450
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 15.67,
|
| 1617 |
+
"learning_rate": 1.0625147440434063e-06,
|
| 1618 |
+
"loss": 0.0706,
|
| 1619 |
+
"step": 2460
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 15.73,
|
| 1623 |
+
"learning_rate": 1.057796650153338e-06,
|
| 1624 |
+
"loss": 0.0688,
|
| 1625 |
+
"step": 2470
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 15.8,
|
| 1629 |
+
"learning_rate": 1.0530785562632695e-06,
|
| 1630 |
+
"loss": 0.0674,
|
| 1631 |
+
"step": 2480
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 15.86,
|
| 1635 |
+
"learning_rate": 1.0483604623732012e-06,
|
| 1636 |
+
"loss": 0.0794,
|
| 1637 |
+
"step": 2490
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 15.92,
|
| 1641 |
+
"learning_rate": 1.0436423684831328e-06,
|
| 1642 |
+
"loss": 0.0604,
|
| 1643 |
+
"step": 2500
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 15.99,
|
| 1647 |
+
"learning_rate": 1.0389242745930644e-06,
|
| 1648 |
+
"loss": 0.0581,
|
| 1649 |
+
"step": 2510
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 16.0,
|
| 1653 |
+
"eval_accuracy": 0.95,
|
| 1654 |
+
"eval_loss": 0.17460142076015472,
|
| 1655 |
+
"eval_runtime": 5.4732,
|
| 1656 |
+
"eval_samples_per_second": 73.083,
|
| 1657 |
+
"eval_steps_per_second": 9.135,
|
| 1658 |
+
"step": 2512
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 16.05,
|
| 1662 |
+
"learning_rate": 1.034206180702996e-06,
|
| 1663 |
+
"loss": 0.0688,
|
| 1664 |
+
"step": 2520
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 16.11,
|
| 1668 |
+
"learning_rate": 1.0294880868129274e-06,
|
| 1669 |
+
"loss": 0.0559,
|
| 1670 |
+
"step": 2530
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 16.18,
|
| 1674 |
+
"learning_rate": 1.024769992922859e-06,
|
| 1675 |
+
"loss": 0.0561,
|
| 1676 |
+
"step": 2540
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 16.24,
|
| 1680 |
+
"learning_rate": 1.0200518990327906e-06,
|
| 1681 |
+
"loss": 0.0558,
|
| 1682 |
+
"step": 2550
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 16.31,
|
| 1686 |
+
"learning_rate": 1.0153338051427223e-06,
|
| 1687 |
+
"loss": 0.0602,
|
| 1688 |
+
"step": 2560
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 16.37,
|
| 1692 |
+
"learning_rate": 1.0106157112526539e-06,
|
| 1693 |
+
"loss": 0.0605,
|
| 1694 |
+
"step": 2570
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 16.43,
|
| 1698 |
+
"learning_rate": 1.0058976173625855e-06,
|
| 1699 |
+
"loss": 0.0516,
|
| 1700 |
+
"step": 2580
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 16.5,
|
| 1704 |
+
"learning_rate": 1.001179523472517e-06,
|
| 1705 |
+
"loss": 0.0566,
|
| 1706 |
+
"step": 2590
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 16.56,
|
| 1710 |
+
"learning_rate": 9.964614295824485e-07,
|
| 1711 |
+
"loss": 0.0624,
|
| 1712 |
+
"step": 2600
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 16.62,
|
| 1716 |
+
"learning_rate": 9.92215145081387e-07,
|
| 1717 |
+
"loss": 0.0522,
|
| 1718 |
+
"step": 2610
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 16.69,
|
| 1722 |
+
"learning_rate": 9.879688605803254e-07,
|
| 1723 |
+
"loss": 0.0563,
|
| 1724 |
+
"step": 2620
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 16.75,
|
| 1728 |
+
"learning_rate": 9.83250766690257e-07,
|
| 1729 |
+
"loss": 0.0583,
|
| 1730 |
+
"step": 2630
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 16.82,
|
| 1734 |
+
"learning_rate": 9.785326728001886e-07,
|
| 1735 |
+
"loss": 0.0587,
|
| 1736 |
+
"step": 2640
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 16.88,
|
| 1740 |
+
"learning_rate": 9.738145789101202e-07,
|
| 1741 |
+
"loss": 0.0552,
|
| 1742 |
+
"step": 2650
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 16.94,
|
| 1746 |
+
"learning_rate": 9.690964850200518e-07,
|
| 1747 |
+
"loss": 0.0612,
|
| 1748 |
+
"step": 2660
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 17.0,
|
| 1752 |
+
"eval_accuracy": 0.9575,
|
| 1753 |
+
"eval_loss": 0.14898547530174255,
|
| 1754 |
+
"eval_runtime": 5.3718,
|
| 1755 |
+
"eval_samples_per_second": 74.462,
|
| 1756 |
+
"eval_steps_per_second": 9.308,
|
| 1757 |
+
"step": 2669
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 17.01,
|
| 1761 |
+
"learning_rate": 9.643783911299834e-07,
|
| 1762 |
+
"loss": 0.0464,
|
| 1763 |
+
"step": 2670
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 17.07,
|
| 1767 |
+
"learning_rate": 9.59660297239915e-07,
|
| 1768 |
+
"loss": 0.049,
|
| 1769 |
+
"step": 2680
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 17.13,
|
| 1773 |
+
"learning_rate": 9.549422033498467e-07,
|
| 1774 |
+
"loss": 0.0484,
|
| 1775 |
+
"step": 2690
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 17.2,
|
| 1779 |
+
"learning_rate": 9.502241094597782e-07,
|
| 1780 |
+
"loss": 0.0419,
|
| 1781 |
+
"step": 2700
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 17.26,
|
| 1785 |
+
"learning_rate": 9.455060155697098e-07,
|
| 1786 |
+
"loss": 0.0422,
|
| 1787 |
+
"step": 2710
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 17.32,
|
| 1791 |
+
"learning_rate": 9.407879216796414e-07,
|
| 1792 |
+
"loss": 0.0476,
|
| 1793 |
+
"step": 2720
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 17.39,
|
| 1797 |
+
"learning_rate": 9.36069827789573e-07,
|
| 1798 |
+
"loss": 0.054,
|
| 1799 |
+
"step": 2730
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 17.45,
|
| 1803 |
+
"learning_rate": 9.313517338995047e-07,
|
| 1804 |
+
"loss": 0.0452,
|
| 1805 |
+
"step": 2740
|
| 1806 |
+
},
|
| 1807 |
+
{
|
| 1808 |
+
"epoch": 17.52,
|
| 1809 |
+
"learning_rate": 9.266336400094362e-07,
|
| 1810 |
+
"loss": 0.0474,
|
| 1811 |
+
"step": 2750
|
| 1812 |
+
},
|
| 1813 |
+
{
|
| 1814 |
+
"epoch": 17.58,
|
| 1815 |
+
"learning_rate": 9.219155461193678e-07,
|
| 1816 |
+
"loss": 0.0406,
|
| 1817 |
+
"step": 2760
|
| 1818 |
+
},
|
| 1819 |
+
{
|
| 1820 |
+
"epoch": 17.64,
|
| 1821 |
+
"learning_rate": 9.171974522292994e-07,
|
| 1822 |
+
"loss": 0.0482,
|
| 1823 |
+
"step": 2770
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 17.71,
|
| 1827 |
+
"learning_rate": 9.124793583392309e-07,
|
| 1828 |
+
"loss": 0.0407,
|
| 1829 |
+
"step": 2780
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 17.77,
|
| 1833 |
+
"learning_rate": 9.077612644491624e-07,
|
| 1834 |
+
"loss": 0.043,
|
| 1835 |
+
"step": 2790
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 17.83,
|
| 1839 |
+
"learning_rate": 9.03043170559094e-07,
|
| 1840 |
+
"loss": 0.0468,
|
| 1841 |
+
"step": 2800
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 17.9,
|
| 1845 |
+
"learning_rate": 8.983250766690256e-07,
|
| 1846 |
+
"loss": 0.0381,
|
| 1847 |
+
"step": 2810
|
| 1848 |
+
},
|
| 1849 |
+
{
|
| 1850 |
+
"epoch": 17.96,
|
| 1851 |
+
"learning_rate": 8.936069827789573e-07,
|
| 1852 |
+
"loss": 0.0515,
|
| 1853 |
+
"step": 2820
|
| 1854 |
+
},
|
| 1855 |
+
{
|
| 1856 |
+
"epoch": 18.0,
|
| 1857 |
+
"eval_accuracy": 0.955,
|
| 1858 |
+
"eval_loss": 0.1454763561487198,
|
| 1859 |
+
"eval_runtime": 5.4502,
|
| 1860 |
+
"eval_samples_per_second": 73.392,
|
| 1861 |
+
"eval_steps_per_second": 9.174,
|
| 1862 |
+
"step": 2826
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 18.03,
|
| 1866 |
+
"learning_rate": 8.888888888888888e-07,
|
| 1867 |
+
"loss": 0.0448,
|
| 1868 |
+
"step": 2830
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 18.09,
|
| 1872 |
+
"learning_rate": 8.841707949988204e-07,
|
| 1873 |
+
"loss": 0.0376,
|
| 1874 |
+
"step": 2840
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 18.15,
|
| 1878 |
+
"learning_rate": 8.79452701108752e-07,
|
| 1879 |
+
"loss": 0.0367,
|
| 1880 |
+
"step": 2850
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 18.22,
|
| 1884 |
+
"learning_rate": 8.747346072186836e-07,
|
| 1885 |
+
"loss": 0.0326,
|
| 1886 |
+
"step": 2860
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 18.28,
|
| 1890 |
+
"learning_rate": 8.700165133286152e-07,
|
| 1891 |
+
"loss": 0.0378,
|
| 1892 |
+
"step": 2870
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 18.34,
|
| 1896 |
+
"learning_rate": 8.652984194385467e-07,
|
| 1897 |
+
"loss": 0.0427,
|
| 1898 |
+
"step": 2880
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 18.41,
|
| 1902 |
+
"learning_rate": 8.605803255484784e-07,
|
| 1903 |
+
"loss": 0.0388,
|
| 1904 |
+
"step": 2890
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 18.47,
|
| 1908 |
+
"learning_rate": 8.5586223165841e-07,
|
| 1909 |
+
"loss": 0.0342,
|
| 1910 |
+
"step": 2900
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 18.54,
|
| 1914 |
+
"learning_rate": 8.511441377683416e-07,
|
| 1915 |
+
"loss": 0.0373,
|
| 1916 |
+
"step": 2910
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 18.6,
|
| 1920 |
+
"learning_rate": 8.464260438782731e-07,
|
| 1921 |
+
"loss": 0.032,
|
| 1922 |
+
"step": 2920
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 18.66,
|
| 1926 |
+
"learning_rate": 8.417079499882047e-07,
|
| 1927 |
+
"loss": 0.0336,
|
| 1928 |
+
"step": 2930
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 18.73,
|
| 1932 |
+
"learning_rate": 8.369898560981363e-07,
|
| 1933 |
+
"loss": 0.0411,
|
| 1934 |
+
"step": 2940
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 18.79,
|
| 1938 |
+
"learning_rate": 8.322717622080679e-07,
|
| 1939 |
+
"loss": 0.0412,
|
| 1940 |
+
"step": 2950
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 18.85,
|
| 1944 |
+
"learning_rate": 8.275536683179994e-07,
|
| 1945 |
+
"loss": 0.036,
|
| 1946 |
+
"step": 2960
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 18.92,
|
| 1950 |
+
"learning_rate": 8.228355744279311e-07,
|
| 1951 |
+
"loss": 0.0319,
|
| 1952 |
+
"step": 2970
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 18.98,
|
| 1956 |
+
"learning_rate": 8.181174805378627e-07,
|
| 1957 |
+
"loss": 0.0339,
|
| 1958 |
+
"step": 2980
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 19.0,
|
| 1962 |
+
"eval_accuracy": 0.965,
|
| 1963 |
+
"eval_loss": 0.13269954919815063,
|
| 1964 |
+
"eval_runtime": 5.908,
|
| 1965 |
+
"eval_samples_per_second": 67.705,
|
| 1966 |
+
"eval_steps_per_second": 8.463,
|
| 1967 |
+
"step": 2983
|
| 1968 |
+
},
|
| 1969 |
+
{
|
| 1970 |
+
"epoch": 19.04,
|
| 1971 |
+
"learning_rate": 8.133993866477943e-07,
|
| 1972 |
+
"loss": 0.0311,
|
| 1973 |
+
"step": 2990
|
| 1974 |
+
},
|
| 1975 |
+
{
|
| 1976 |
+
"epoch": 19.11,
|
| 1977 |
+
"learning_rate": 8.086812927577259e-07,
|
| 1978 |
+
"loss": 0.0308,
|
| 1979 |
+
"step": 3000
|
| 1980 |
+
},
|
| 1981 |
+
{
|
| 1982 |
+
"epoch": 19.17,
|
| 1983 |
+
"learning_rate": 8.039631988676574e-07,
|
| 1984 |
+
"loss": 0.0285,
|
| 1985 |
+
"step": 3010
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 19.24,
|
| 1989 |
+
"learning_rate": 7.99245104977589e-07,
|
| 1990 |
+
"loss": 0.0307,
|
| 1991 |
+
"step": 3020
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 19.3,
|
| 1995 |
+
"learning_rate": 7.945270110875207e-07,
|
| 1996 |
+
"loss": 0.0332,
|
| 1997 |
+
"step": 3030
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 19.36,
|
| 2001 |
+
"learning_rate": 7.898089171974523e-07,
|
| 2002 |
+
"loss": 0.0334,
|
| 2003 |
+
"step": 3040
|
| 2004 |
+
},
|
| 2005 |
+
{
|
| 2006 |
+
"epoch": 19.43,
|
| 2007 |
+
"learning_rate": 7.850908233073838e-07,
|
| 2008 |
+
"loss": 0.0337,
|
| 2009 |
+
"step": 3050
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 19.49,
|
| 2013 |
+
"learning_rate": 7.803727294173154e-07,
|
| 2014 |
+
"loss": 0.0254,
|
| 2015 |
+
"step": 3060
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 19.55,
|
| 2019 |
+
"learning_rate": 7.75654635527247e-07,
|
| 2020 |
+
"loss": 0.0272,
|
| 2021 |
+
"step": 3070
|
| 2022 |
+
},
|
| 2023 |
+
{
|
| 2024 |
+
"epoch": 19.62,
|
| 2025 |
+
"learning_rate": 7.709365416371786e-07,
|
| 2026 |
+
"loss": 0.0256,
|
| 2027 |
+
"step": 3080
|
| 2028 |
+
},
|
| 2029 |
+
{
|
| 2030 |
+
"epoch": 19.68,
|
| 2031 |
+
"learning_rate": 7.662184477471101e-07,
|
| 2032 |
+
"loss": 0.0334,
|
| 2033 |
+
"step": 3090
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 19.75,
|
| 2037 |
+
"learning_rate": 7.615003538570417e-07,
|
| 2038 |
+
"loss": 0.0302,
|
| 2039 |
+
"step": 3100
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 19.81,
|
| 2043 |
+
"learning_rate": 7.567822599669734e-07,
|
| 2044 |
+
"loss": 0.0243,
|
| 2045 |
+
"step": 3110
|
| 2046 |
+
},
|
| 2047 |
+
{
|
| 2048 |
+
"epoch": 19.87,
|
| 2049 |
+
"learning_rate": 7.52064166076905e-07,
|
| 2050 |
+
"loss": 0.0284,
|
| 2051 |
+
"step": 3120
|
| 2052 |
+
},
|
| 2053 |
+
{
|
| 2054 |
+
"epoch": 19.94,
|
| 2055 |
+
"learning_rate": 7.473460721868364e-07,
|
| 2056 |
+
"loss": 0.0291,
|
| 2057 |
+
"step": 3130
|
| 2058 |
+
},
|
| 2059 |
+
{
|
| 2060 |
+
"epoch": 20.0,
|
| 2061 |
+
"learning_rate": 7.42627978296768e-07,
|
| 2062 |
+
"loss": 0.0251,
|
| 2063 |
+
"step": 3140
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 20.0,
|
| 2067 |
+
"eval_accuracy": 0.955,
|
| 2068 |
+
"eval_loss": 0.13997198641300201,
|
| 2069 |
+
"eval_runtime": 6.1227,
|
| 2070 |
+
"eval_samples_per_second": 65.33,
|
| 2071 |
+
"eval_steps_per_second": 8.166,
|
| 2072 |
+
"step": 3140
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 20.06,
|
| 2076 |
+
"learning_rate": 7.379098844066996e-07,
|
| 2077 |
+
"loss": 0.026,
|
| 2078 |
+
"step": 3150
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 20.13,
|
| 2082 |
+
"learning_rate": 7.331917905166312e-07,
|
| 2083 |
+
"loss": 0.0243,
|
| 2084 |
+
"step": 3160
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 20.19,
|
| 2088 |
+
"learning_rate": 7.284736966265628e-07,
|
| 2089 |
+
"loss": 0.0299,
|
| 2090 |
+
"step": 3170
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 20.25,
|
| 2094 |
+
"learning_rate": 7.237556027364944e-07,
|
| 2095 |
+
"loss": 0.0253,
|
| 2096 |
+
"step": 3180
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 20.32,
|
| 2100 |
+
"learning_rate": 7.19037508846426e-07,
|
| 2101 |
+
"loss": 0.0266,
|
| 2102 |
+
"step": 3190
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 20.38,
|
| 2106 |
+
"learning_rate": 7.143194149563576e-07,
|
| 2107 |
+
"loss": 0.0227,
|
| 2108 |
+
"step": 3200
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 20.45,
|
| 2112 |
+
"learning_rate": 7.096013210662892e-07,
|
| 2113 |
+
"loss": 0.0227,
|
| 2114 |
+
"step": 3210
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 20.51,
|
| 2118 |
+
"learning_rate": 7.048832271762207e-07,
|
| 2119 |
+
"loss": 0.0234,
|
| 2120 |
+
"step": 3220
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 20.57,
|
| 2124 |
+
"learning_rate": 7.001651332861523e-07,
|
| 2125 |
+
"loss": 0.0238,
|
| 2126 |
+
"step": 3230
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 20.64,
|
| 2130 |
+
"learning_rate": 6.954470393960839e-07,
|
| 2131 |
+
"loss": 0.0278,
|
| 2132 |
+
"step": 3240
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 20.7,
|
| 2136 |
+
"learning_rate": 6.907289455060156e-07,
|
| 2137 |
+
"loss": 0.0211,
|
| 2138 |
+
"step": 3250
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 20.76,
|
| 2142 |
+
"learning_rate": 6.860108516159471e-07,
|
| 2143 |
+
"loss": 0.0196,
|
| 2144 |
+
"step": 3260
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 20.83,
|
| 2148 |
+
"learning_rate": 6.812927577258787e-07,
|
| 2149 |
+
"loss": 0.019,
|
| 2150 |
+
"step": 3270
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 20.89,
|
| 2154 |
+
"learning_rate": 6.765746638358103e-07,
|
| 2155 |
+
"loss": 0.0213,
|
| 2156 |
+
"step": 3280
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 20.96,
|
| 2160 |
+
"learning_rate": 6.718565699457419e-07,
|
| 2161 |
+
"loss": 0.0216,
|
| 2162 |
+
"step": 3290
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 21.0,
|
| 2166 |
+
"eval_accuracy": 0.96,
|
| 2167 |
+
"eval_loss": 0.13987915217876434,
|
| 2168 |
+
"eval_runtime": 6.4007,
|
| 2169 |
+
"eval_samples_per_second": 62.494,
|
| 2170 |
+
"eval_steps_per_second": 7.812,
|
| 2171 |
+
"step": 3297
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 21.02,
|
| 2175 |
+
"learning_rate": 6.671384760556735e-07,
|
| 2176 |
+
"loss": 0.023,
|
| 2177 |
+
"step": 3300
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 21.08,
|
| 2181 |
+
"learning_rate": 6.62420382165605e-07,
|
| 2182 |
+
"loss": 0.022,
|
| 2183 |
+
"step": 3310
|
| 2184 |
+
},
|
| 2185 |
+
{
|
| 2186 |
+
"epoch": 21.15,
|
| 2187 |
+
"learning_rate": 6.577022882755366e-07,
|
| 2188 |
+
"loss": 0.02,
|
| 2189 |
+
"step": 3320
|
| 2190 |
+
},
|
| 2191 |
+
{
|
| 2192 |
+
"epoch": 21.21,
|
| 2193 |
+
"learning_rate": 6.529841943854683e-07,
|
| 2194 |
+
"loss": 0.019,
|
| 2195 |
+
"step": 3330
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 21.27,
|
| 2199 |
+
"learning_rate": 6.482661004953999e-07,
|
| 2200 |
+
"loss": 0.0188,
|
| 2201 |
+
"step": 3340
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 21.34,
|
| 2205 |
+
"learning_rate": 6.435480066053314e-07,
|
| 2206 |
+
"loss": 0.0186,
|
| 2207 |
+
"step": 3350
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 21.4,
|
| 2211 |
+
"learning_rate": 6.38829912715263e-07,
|
| 2212 |
+
"loss": 0.0182,
|
| 2213 |
+
"step": 3360
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 21.46,
|
| 2217 |
+
"learning_rate": 6.341118188251946e-07,
|
| 2218 |
+
"loss": 0.0182,
|
| 2219 |
+
"step": 3370
|
| 2220 |
+
},
|
| 2221 |
+
{
|
| 2222 |
+
"epoch": 21.53,
|
| 2223 |
+
"learning_rate": 6.293937249351262e-07,
|
| 2224 |
+
"loss": 0.0233,
|
| 2225 |
+
"step": 3380
|
| 2226 |
+
},
|
| 2227 |
+
{
|
| 2228 |
+
"epoch": 21.59,
|
| 2229 |
+
"learning_rate": 6.246756310450577e-07,
|
| 2230 |
+
"loss": 0.0234,
|
| 2231 |
+
"step": 3390
|
| 2232 |
+
},
|
| 2233 |
+
{
|
| 2234 |
+
"epoch": 21.66,
|
| 2235 |
+
"learning_rate": 6.199575371549894e-07,
|
| 2236 |
+
"loss": 0.0213,
|
| 2237 |
+
"step": 3400
|
| 2238 |
+
},
|
| 2239 |
+
{
|
| 2240 |
+
"epoch": 21.72,
|
| 2241 |
+
"learning_rate": 6.15239443264921e-07,
|
| 2242 |
+
"loss": 0.0226,
|
| 2243 |
+
"step": 3410
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 21.78,
|
| 2247 |
+
"learning_rate": 6.105213493748526e-07,
|
| 2248 |
+
"loss": 0.016,
|
| 2249 |
+
"step": 3420
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 21.85,
|
| 2253 |
+
"learning_rate": 6.058032554847842e-07,
|
| 2254 |
+
"loss": 0.0177,
|
| 2255 |
+
"step": 3430
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 21.91,
|
| 2259 |
+
"learning_rate": 6.010851615947157e-07,
|
| 2260 |
+
"loss": 0.0183,
|
| 2261 |
+
"step": 3440
|
| 2262 |
+
},
|
| 2263 |
+
{
|
| 2264 |
+
"epoch": 21.97,
|
| 2265 |
+
"learning_rate": 5.963670677046473e-07,
|
| 2266 |
+
"loss": 0.0159,
|
| 2267 |
+
"step": 3450
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 22.0,
|
| 2271 |
+
"eval_accuracy": 0.96,
|
| 2272 |
+
"eval_loss": 0.12578421831130981,
|
| 2273 |
+
"eval_runtime": 6.9215,
|
| 2274 |
+
"eval_samples_per_second": 57.791,
|
| 2275 |
+
"eval_steps_per_second": 7.224,
|
| 2276 |
+
"step": 3454
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 22.04,
|
| 2280 |
+
"learning_rate": 5.916489738145789e-07,
|
| 2281 |
+
"loss": 0.0183,
|
| 2282 |
+
"step": 3460
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 22.1,
|
| 2286 |
+
"learning_rate": 5.869308799245106e-07,
|
| 2287 |
+
"loss": 0.0182,
|
| 2288 |
+
"step": 3470
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 22.17,
|
| 2292 |
+
"learning_rate": 5.82212786034442e-07,
|
| 2293 |
+
"loss": 0.0163,
|
| 2294 |
+
"step": 3480
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 22.23,
|
| 2298 |
+
"learning_rate": 5.774946921443736e-07,
|
| 2299 |
+
"loss": 0.0193,
|
| 2300 |
+
"step": 3490
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"epoch": 22.29,
|
| 2304 |
+
"learning_rate": 5.727765982543052e-07,
|
| 2305 |
+
"loss": 0.0176,
|
| 2306 |
+
"step": 3500
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 22.36,
|
| 2310 |
+
"learning_rate": 5.680585043642368e-07,
|
| 2311 |
+
"loss": 0.0156,
|
| 2312 |
+
"step": 3510
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 22.42,
|
| 2316 |
+
"learning_rate": 5.633404104741683e-07,
|
| 2317 |
+
"loss": 0.016,
|
| 2318 |
+
"step": 3520
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 22.48,
|
| 2322 |
+
"learning_rate": 5.586223165840999e-07,
|
| 2323 |
+
"loss": 0.015,
|
| 2324 |
+
"step": 3530
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 22.55,
|
| 2328 |
+
"learning_rate": 5.539042226940316e-07,
|
| 2329 |
+
"loss": 0.0166,
|
| 2330 |
+
"step": 3540
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 22.61,
|
| 2334 |
+
"learning_rate": 5.491861288039632e-07,
|
| 2335 |
+
"loss": 0.0156,
|
| 2336 |
+
"step": 3550
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 22.68,
|
| 2340 |
+
"learning_rate": 5.444680349138948e-07,
|
| 2341 |
+
"loss": 0.0152,
|
| 2342 |
+
"step": 3560
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 22.74,
|
| 2346 |
+
"learning_rate": 5.397499410238263e-07,
|
| 2347 |
+
"loss": 0.0179,
|
| 2348 |
+
"step": 3570
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 22.8,
|
| 2352 |
+
"learning_rate": 5.350318471337579e-07,
|
| 2353 |
+
"loss": 0.0148,
|
| 2354 |
+
"step": 3580
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 22.87,
|
| 2358 |
+
"learning_rate": 5.303137532436895e-07,
|
| 2359 |
+
"loss": 0.0157,
|
| 2360 |
+
"step": 3590
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 22.93,
|
| 2364 |
+
"learning_rate": 5.255956593536211e-07,
|
| 2365 |
+
"loss": 0.016,
|
| 2366 |
+
"step": 3600
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 22.99,
|
| 2370 |
+
"learning_rate": 5.208775654635526e-07,
|
| 2371 |
+
"loss": 0.0137,
|
| 2372 |
+
"step": 3610
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 23.0,
|
| 2376 |
+
"eval_accuracy": 0.9625,
|
| 2377 |
+
"eval_loss": 0.12316268682479858,
|
| 2378 |
+
"eval_runtime": 6.7883,
|
| 2379 |
+
"eval_samples_per_second": 58.925,
|
| 2380 |
+
"eval_steps_per_second": 7.366,
|
| 2381 |
+
"step": 3611
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 23.06,
|
| 2385 |
+
"learning_rate": 5.161594715734843e-07,
|
| 2386 |
+
"loss": 0.0141,
|
| 2387 |
+
"step": 3620
|
| 2388 |
+
},
|
| 2389 |
+
{
|
| 2390 |
+
"epoch": 23.12,
|
| 2391 |
+
"learning_rate": 5.114413776834159e-07,
|
| 2392 |
+
"loss": 0.0116,
|
| 2393 |
+
"step": 3630
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 23.18,
|
| 2397 |
+
"learning_rate": 5.067232837933475e-07,
|
| 2398 |
+
"loss": 0.0142,
|
| 2399 |
+
"step": 3640
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 23.25,
|
| 2403 |
+
"learning_rate": 5.02005189903279e-07,
|
| 2404 |
+
"loss": 0.0126,
|
| 2405 |
+
"step": 3650
|
| 2406 |
+
},
|
| 2407 |
+
{
|
| 2408 |
+
"epoch": 23.31,
|
| 2409 |
+
"learning_rate": 4.972870960132106e-07,
|
| 2410 |
+
"loss": 0.0138,
|
| 2411 |
+
"step": 3660
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 23.38,
|
| 2415 |
+
"learning_rate": 4.925690021231422e-07,
|
| 2416 |
+
"loss": 0.0135,
|
| 2417 |
+
"step": 3670
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 23.44,
|
| 2421 |
+
"learning_rate": 4.878509082330738e-07,
|
| 2422 |
+
"loss": 0.0144,
|
| 2423 |
+
"step": 3680
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 23.5,
|
| 2427 |
+
"learning_rate": 4.831328143430054e-07,
|
| 2428 |
+
"loss": 0.0147,
|
| 2429 |
+
"step": 3690
|
| 2430 |
+
},
|
| 2431 |
+
{
|
| 2432 |
+
"epoch": 23.57,
|
| 2433 |
+
"learning_rate": 4.78414720452937e-07,
|
| 2434 |
+
"loss": 0.0131,
|
| 2435 |
+
"step": 3700
|
| 2436 |
+
},
|
| 2437 |
+
{
|
| 2438 |
+
"epoch": 23.63,
|
| 2439 |
+
"learning_rate": 4.736966265628686e-07,
|
| 2440 |
+
"loss": 0.0142,
|
| 2441 |
+
"step": 3710
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 23.69,
|
| 2445 |
+
"learning_rate": 4.689785326728002e-07,
|
| 2446 |
+
"loss": 0.0122,
|
| 2447 |
+
"step": 3720
|
| 2448 |
+
},
|
| 2449 |
+
{
|
| 2450 |
+
"epoch": 23.76,
|
| 2451 |
+
"learning_rate": 4.6426043878273177e-07,
|
| 2452 |
+
"loss": 0.0179,
|
| 2453 |
+
"step": 3730
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 23.82,
|
| 2457 |
+
"learning_rate": 4.595423448926634e-07,
|
| 2458 |
+
"loss": 0.0152,
|
| 2459 |
+
"step": 3740
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 23.89,
|
| 2463 |
+
"learning_rate": 4.548242510025949e-07,
|
| 2464 |
+
"loss": 0.015,
|
| 2465 |
+
"step": 3750
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 23.95,
|
| 2469 |
+
"learning_rate": 4.501061571125265e-07,
|
| 2470 |
+
"loss": 0.0135,
|
| 2471 |
+
"step": 3760
|
| 2472 |
+
},
|
| 2473 |
+
{
|
| 2474 |
+
"epoch": 24.0,
|
| 2475 |
+
"eval_accuracy": 0.97,
|
| 2476 |
+
"eval_loss": 0.12535163760185242,
|
| 2477 |
+
"eval_runtime": 6.4232,
|
| 2478 |
+
"eval_samples_per_second": 62.274,
|
| 2479 |
+
"eval_steps_per_second": 7.784,
|
| 2480 |
+
"step": 3768
|
| 2481 |
+
}
|
| 2482 |
+
],
|
| 2483 |
+
"max_steps": 4710,
|
| 2484 |
+
"num_train_epochs": 30,
|
| 2485 |
+
"total_flos": 1.859841088487424e+19,
|
| 2486 |
+
"trial_name": null,
|
| 2487 |
+
"trial_params": null
|
| 2488 |
+
}
|
checkpoint-3768/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9259cd92d8f5e48e51c9df15d7cdc796908ee099294f7d0057b13d585a7eda0d
|
| 3 |
+
size 3579
|