Upload 8 files
Browse files- config.json +30 -0
- optimizer.pt +3 -0
- preprocessor_config.json +22 -0
- pytorch_model.bin +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +1741 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/vit-base-patch16-224-in21k",
|
| 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": "LABEL_0"
|
| 13 |
+
},
|
| 14 |
+
"image_size": 224,
|
| 15 |
+
"initializer_range": 0.02,
|
| 16 |
+
"intermediate_size": 3072,
|
| 17 |
+
"label2id": {
|
| 18 |
+
"LABEL_0": 0
|
| 19 |
+
},
|
| 20 |
+
"layer_norm_eps": 1e-12,
|
| 21 |
+
"model_type": "vit",
|
| 22 |
+
"num_attention_heads": 12,
|
| 23 |
+
"num_channels": 3,
|
| 24 |
+
"num_hidden_layers": 12,
|
| 25 |
+
"patch_size": 16,
|
| 26 |
+
"problem_type": "regression",
|
| 27 |
+
"qkv_bias": true,
|
| 28 |
+
"torch_dtype": "float32",
|
| 29 |
+
"transformers_version": "4.29.1"
|
| 30 |
+
}
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d43282583f4e6291a3d40f788104fb32d3d5fe3a8ad3d69eaac9f5bcbd07f652
|
| 3 |
+
size 686512773
|
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": "ViTFeatureExtractor",
|
| 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 |
+
}
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cd1776ccb0275a483d9293c5a4cd5eda00e2efe35feb16677adda2460e74762a
|
| 3 |
+
size 343265645
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4a2d2d1561481739972beb9da4cda24cd79618e56b2904d90829eda87023314e
|
| 3 |
+
size 14575
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2fff95da4b494ce7c23f35e63d158ff7b0ff3b14cced25a08f2bf7202f00e540
|
| 3 |
+
size 627
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1741 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 108.29063415527344,
|
| 3 |
+
"best_model_checkpoint": "./ViT_ordinary/checkpoint-2500",
|
| 4 |
+
"epoch": 2.2893772893772892,
|
| 5 |
+
"global_step": 2500,
|
| 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": 1.1452785890167783e-06,
|
| 13 |
+
"loss": 1143.4345,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.02,
|
| 18 |
+
"learning_rate": 2.2905571780335565e-06,
|
| 19 |
+
"loss": 1227.331,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.03,
|
| 24 |
+
"learning_rate": 3.435835767050335e-06,
|
| 25 |
+
"loss": 1257.7006,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.04,
|
| 30 |
+
"learning_rate": 4.581114356067113e-06,
|
| 31 |
+
"loss": 1220.7573,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.05,
|
| 36 |
+
"learning_rate": 5.726392945083892e-06,
|
| 37 |
+
"loss": 1314.9278,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.05,
|
| 42 |
+
"learning_rate": 6.87167153410067e-06,
|
| 43 |
+
"loss": 1161.0052,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.06,
|
| 48 |
+
"learning_rate": 8.016950123117448e-06,
|
| 49 |
+
"loss": 1195.8272,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.07,
|
| 54 |
+
"learning_rate": 9.162228712134226e-06,
|
| 55 |
+
"loss": 1135.2035,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.08,
|
| 60 |
+
"learning_rate": 1.0307507301151004e-05,
|
| 61 |
+
"loss": 1105.4266,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.09,
|
| 66 |
+
"learning_rate": 1.1452785890167784e-05,
|
| 67 |
+
"loss": 1138.5144,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.09,
|
| 72 |
+
"eval_RMSE": 45.47388458251953,
|
| 73 |
+
"eval_loss": 2676.440673828125,
|
| 74 |
+
"eval_runtime": 88.8721,
|
| 75 |
+
"eval_samples_per_second": 91.941,
|
| 76 |
+
"eval_steps_per_second": 2.881,
|
| 77 |
+
"step": 100
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 0.1,
|
| 81 |
+
"learning_rate": 1.259806447918456e-05,
|
| 82 |
+
"loss": 936.6518,
|
| 83 |
+
"step": 110
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.11,
|
| 87 |
+
"learning_rate": 1.374334306820134e-05,
|
| 88 |
+
"loss": 1003.3095,
|
| 89 |
+
"step": 120
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.12,
|
| 93 |
+
"learning_rate": 1.4888621657218118e-05,
|
| 94 |
+
"loss": 994.7256,
|
| 95 |
+
"step": 130
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 0.13,
|
| 99 |
+
"learning_rate": 1.6033900246234896e-05,
|
| 100 |
+
"loss": 962.0123,
|
| 101 |
+
"step": 140
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.14,
|
| 105 |
+
"learning_rate": 1.7179178835251674e-05,
|
| 106 |
+
"loss": 1012.3558,
|
| 107 |
+
"step": 150
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.15,
|
| 111 |
+
"learning_rate": 1.8324457424268452e-05,
|
| 112 |
+
"loss": 923.0986,
|
| 113 |
+
"step": 160
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 0.16,
|
| 117 |
+
"learning_rate": 1.946973601328523e-05,
|
| 118 |
+
"loss": 1055.8317,
|
| 119 |
+
"step": 170
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 0.16,
|
| 123 |
+
"learning_rate": 2.0615014602302008e-05,
|
| 124 |
+
"loss": 1189.4732,
|
| 125 |
+
"step": 180
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.17,
|
| 129 |
+
"learning_rate": 2.1760293191318786e-05,
|
| 130 |
+
"loss": 998.1689,
|
| 131 |
+
"step": 190
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 0.18,
|
| 135 |
+
"learning_rate": 2.2905571780335568e-05,
|
| 136 |
+
"loss": 940.4191,
|
| 137 |
+
"step": 200
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.18,
|
| 141 |
+
"eval_RMSE": 44.35804748535156,
|
| 142 |
+
"eval_loss": 2574.700927734375,
|
| 143 |
+
"eval_runtime": 88.6183,
|
| 144 |
+
"eval_samples_per_second": 92.204,
|
| 145 |
+
"eval_steps_per_second": 2.889,
|
| 146 |
+
"step": 200
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.19,
|
| 150 |
+
"learning_rate": 2.4050850369352342e-05,
|
| 151 |
+
"loss": 1081.1541,
|
| 152 |
+
"step": 210
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.2,
|
| 156 |
+
"learning_rate": 2.519612895836912e-05,
|
| 157 |
+
"loss": 953.7207,
|
| 158 |
+
"step": 220
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.21,
|
| 162 |
+
"learning_rate": 2.63414075473859e-05,
|
| 163 |
+
"loss": 920.4622,
|
| 164 |
+
"step": 230
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.22,
|
| 168 |
+
"learning_rate": 2.748668613640268e-05,
|
| 169 |
+
"loss": 1037.0487,
|
| 170 |
+
"step": 240
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.23,
|
| 174 |
+
"learning_rate": 2.8631964725419458e-05,
|
| 175 |
+
"loss": 991.3824,
|
| 176 |
+
"step": 250
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.24,
|
| 180 |
+
"learning_rate": 2.9777243314436236e-05,
|
| 181 |
+
"loss": 1079.5839,
|
| 182 |
+
"step": 260
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.25,
|
| 186 |
+
"learning_rate": 3.092252190345301e-05,
|
| 187 |
+
"loss": 989.4474,
|
| 188 |
+
"step": 270
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.26,
|
| 192 |
+
"learning_rate": 3.206780049246979e-05,
|
| 193 |
+
"loss": 987.4904,
|
| 194 |
+
"step": 280
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.27,
|
| 198 |
+
"learning_rate": 3.321307908148657e-05,
|
| 199 |
+
"loss": 1038.7738,
|
| 200 |
+
"step": 290
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.27,
|
| 204 |
+
"learning_rate": 3.435835767050335e-05,
|
| 205 |
+
"loss": 921.6753,
|
| 206 |
+
"step": 300
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.27,
|
| 210 |
+
"eval_RMSE": 43.82501220703125,
|
| 211 |
+
"eval_loss": 2526.614501953125,
|
| 212 |
+
"eval_runtime": 89.1648,
|
| 213 |
+
"eval_samples_per_second": 91.639,
|
| 214 |
+
"eval_steps_per_second": 2.871,
|
| 215 |
+
"step": 300
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.28,
|
| 219 |
+
"learning_rate": 3.550363625952013e-05,
|
| 220 |
+
"loss": 909.3841,
|
| 221 |
+
"step": 310
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.29,
|
| 225 |
+
"learning_rate": 3.6648914848536904e-05,
|
| 226 |
+
"loss": 931.0681,
|
| 227 |
+
"step": 320
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.3,
|
| 231 |
+
"learning_rate": 3.7794193437553686e-05,
|
| 232 |
+
"loss": 969.0332,
|
| 233 |
+
"step": 330
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.31,
|
| 237 |
+
"learning_rate": 3.893947202657046e-05,
|
| 238 |
+
"loss": 1054.9324,
|
| 239 |
+
"step": 340
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 0.32,
|
| 243 |
+
"learning_rate": 4.008475061558724e-05,
|
| 244 |
+
"loss": 989.6485,
|
| 245 |
+
"step": 350
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 0.33,
|
| 249 |
+
"learning_rate": 4.1230029204604016e-05,
|
| 250 |
+
"loss": 940.8793,
|
| 251 |
+
"step": 360
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.34,
|
| 255 |
+
"learning_rate": 4.237530779362079e-05,
|
| 256 |
+
"loss": 935.6209,
|
| 257 |
+
"step": 370
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.35,
|
| 261 |
+
"learning_rate": 4.352058638263757e-05,
|
| 262 |
+
"loss": 1083.344,
|
| 263 |
+
"step": 380
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.36,
|
| 267 |
+
"learning_rate": 4.4665864971654354e-05,
|
| 268 |
+
"loss": 918.0279,
|
| 269 |
+
"step": 390
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.37,
|
| 273 |
+
"learning_rate": 4.5811143560671135e-05,
|
| 274 |
+
"loss": 938.6592,
|
| 275 |
+
"step": 400
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.37,
|
| 279 |
+
"eval_RMSE": 43.11045837402344,
|
| 280 |
+
"eval_loss": 2458.26171875,
|
| 281 |
+
"eval_runtime": 89.3306,
|
| 282 |
+
"eval_samples_per_second": 91.469,
|
| 283 |
+
"eval_steps_per_second": 2.866,
|
| 284 |
+
"step": 400
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.38,
|
| 288 |
+
"learning_rate": 4.69564221496879e-05,
|
| 289 |
+
"loss": 980.8854,
|
| 290 |
+
"step": 410
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.38,
|
| 294 |
+
"learning_rate": 4.8101700738704685e-05,
|
| 295 |
+
"loss": 1039.0189,
|
| 296 |
+
"step": 420
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.39,
|
| 300 |
+
"learning_rate": 4.9246979327721466e-05,
|
| 301 |
+
"loss": 876.2118,
|
| 302 |
+
"step": 430
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.4,
|
| 306 |
+
"learning_rate": 5.039225791673824e-05,
|
| 307 |
+
"loss": 922.5582,
|
| 308 |
+
"step": 440
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.41,
|
| 312 |
+
"learning_rate": 5.153753650575502e-05,
|
| 313 |
+
"loss": 918.9763,
|
| 314 |
+
"step": 450
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.42,
|
| 318 |
+
"learning_rate": 5.26828150947718e-05,
|
| 319 |
+
"loss": 967.343,
|
| 320 |
+
"step": 460
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.43,
|
| 324 |
+
"learning_rate": 5.3828093683788585e-05,
|
| 325 |
+
"loss": 866.294,
|
| 326 |
+
"step": 470
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.44,
|
| 330 |
+
"learning_rate": 5.497337227280536e-05,
|
| 331 |
+
"loss": 866.6839,
|
| 332 |
+
"step": 480
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.45,
|
| 336 |
+
"learning_rate": 5.611865086182213e-05,
|
| 337 |
+
"loss": 839.3744,
|
| 338 |
+
"step": 490
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.46,
|
| 342 |
+
"learning_rate": 5.7263929450838916e-05,
|
| 343 |
+
"loss": 904.052,
|
| 344 |
+
"step": 500
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.46,
|
| 348 |
+
"eval_RMSE": 42.12101364135742,
|
| 349 |
+
"eval_loss": 2360.75439453125,
|
| 350 |
+
"eval_runtime": 82.5537,
|
| 351 |
+
"eval_samples_per_second": 98.978,
|
| 352 |
+
"eval_steps_per_second": 3.101,
|
| 353 |
+
"step": 500
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.47,
|
| 357 |
+
"learning_rate": 5.840920803985569e-05,
|
| 358 |
+
"loss": 770.1921,
|
| 359 |
+
"step": 510
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.48,
|
| 363 |
+
"learning_rate": 5.955448662887247e-05,
|
| 364 |
+
"loss": 867.2703,
|
| 365 |
+
"step": 520
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 0.49,
|
| 369 |
+
"learning_rate": 6.0699765217889246e-05,
|
| 370 |
+
"loss": 803.1129,
|
| 371 |
+
"step": 530
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.49,
|
| 375 |
+
"learning_rate": 6.184504380690602e-05,
|
| 376 |
+
"loss": 838.8234,
|
| 377 |
+
"step": 540
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.5,
|
| 381 |
+
"learning_rate": 6.29903223959228e-05,
|
| 382 |
+
"loss": 854.0321,
|
| 383 |
+
"step": 550
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 0.51,
|
| 387 |
+
"learning_rate": 6.413560098493958e-05,
|
| 388 |
+
"loss": 843.9535,
|
| 389 |
+
"step": 560
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.52,
|
| 393 |
+
"learning_rate": 6.528087957395637e-05,
|
| 394 |
+
"loss": 819.4999,
|
| 395 |
+
"step": 570
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.53,
|
| 399 |
+
"learning_rate": 6.642615816297315e-05,
|
| 400 |
+
"loss": 833.5234,
|
| 401 |
+
"step": 580
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.54,
|
| 405 |
+
"learning_rate": 6.757143675198991e-05,
|
| 406 |
+
"loss": 919.277,
|
| 407 |
+
"step": 590
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 0.55,
|
| 411 |
+
"learning_rate": 6.87167153410067e-05,
|
| 412 |
+
"loss": 852.848,
|
| 413 |
+
"step": 600
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.55,
|
| 417 |
+
"eval_RMSE": 40.79917526245117,
|
| 418 |
+
"eval_loss": 2225.54150390625,
|
| 419 |
+
"eval_runtime": 90.1848,
|
| 420 |
+
"eval_samples_per_second": 90.603,
|
| 421 |
+
"eval_steps_per_second": 2.839,
|
| 422 |
+
"step": 600
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.56,
|
| 426 |
+
"learning_rate": 6.986199393002348e-05,
|
| 427 |
+
"loss": 750.3481,
|
| 428 |
+
"step": 610
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.57,
|
| 432 |
+
"learning_rate": 7.100727251904026e-05,
|
| 433 |
+
"loss": 879.5618,
|
| 434 |
+
"step": 620
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.58,
|
| 438 |
+
"learning_rate": 7.215255110805703e-05,
|
| 439 |
+
"loss": 797.6468,
|
| 440 |
+
"step": 630
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 0.59,
|
| 444 |
+
"learning_rate": 7.329782969707381e-05,
|
| 445 |
+
"loss": 835.0447,
|
| 446 |
+
"step": 640
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.6,
|
| 450 |
+
"learning_rate": 7.444310828609059e-05,
|
| 451 |
+
"loss": 755.9302,
|
| 452 |
+
"step": 650
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.6,
|
| 456 |
+
"learning_rate": 7.558838687510737e-05,
|
| 457 |
+
"loss": 749.4049,
|
| 458 |
+
"step": 660
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.61,
|
| 462 |
+
"learning_rate": 7.673366546412415e-05,
|
| 463 |
+
"loss": 727.4822,
|
| 464 |
+
"step": 670
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.62,
|
| 468 |
+
"learning_rate": 7.787894405314092e-05,
|
| 469 |
+
"loss": 664.3502,
|
| 470 |
+
"step": 680
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.63,
|
| 474 |
+
"learning_rate": 7.90242226421577e-05,
|
| 475 |
+
"loss": 737.5342,
|
| 476 |
+
"step": 690
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.64,
|
| 480 |
+
"learning_rate": 8.016950123117448e-05,
|
| 481 |
+
"loss": 685.9979,
|
| 482 |
+
"step": 700
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.64,
|
| 486 |
+
"eval_RMSE": 39.182891845703125,
|
| 487 |
+
"eval_loss": 2045.955322265625,
|
| 488 |
+
"eval_runtime": 86.8947,
|
| 489 |
+
"eval_samples_per_second": 94.033,
|
| 490 |
+
"eval_steps_per_second": 2.946,
|
| 491 |
+
"step": 700
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.65,
|
| 495 |
+
"learning_rate": 8.131477982019125e-05,
|
| 496 |
+
"loss": 818.1389,
|
| 497 |
+
"step": 710
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 0.66,
|
| 501 |
+
"learning_rate": 8.246005840920803e-05,
|
| 502 |
+
"loss": 679.9854,
|
| 503 |
+
"step": 720
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 0.67,
|
| 507 |
+
"learning_rate": 8.360533699822481e-05,
|
| 508 |
+
"loss": 696.1048,
|
| 509 |
+
"step": 730
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 0.68,
|
| 513 |
+
"learning_rate": 8.475061558724158e-05,
|
| 514 |
+
"loss": 694.8774,
|
| 515 |
+
"step": 740
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.69,
|
| 519 |
+
"learning_rate": 8.589589417625838e-05,
|
| 520 |
+
"loss": 673.0095,
|
| 521 |
+
"step": 750
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.7,
|
| 525 |
+
"learning_rate": 8.704117276527514e-05,
|
| 526 |
+
"loss": 638.9566,
|
| 527 |
+
"step": 760
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.71,
|
| 531 |
+
"learning_rate": 8.818645135429193e-05,
|
| 532 |
+
"loss": 660.6367,
|
| 533 |
+
"step": 770
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.71,
|
| 537 |
+
"learning_rate": 8.933172994330871e-05,
|
| 538 |
+
"loss": 622.8923,
|
| 539 |
+
"step": 780
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.72,
|
| 543 |
+
"learning_rate": 9.047700853232548e-05,
|
| 544 |
+
"loss": 570.6501,
|
| 545 |
+
"step": 790
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 0.73,
|
| 549 |
+
"learning_rate": 9.162228712134227e-05,
|
| 550 |
+
"loss": 612.6461,
|
| 551 |
+
"step": 800
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 0.73,
|
| 555 |
+
"eval_RMSE": 37.14878845214844,
|
| 556 |
+
"eval_loss": 1821.157958984375,
|
| 557 |
+
"eval_runtime": 90.6758,
|
| 558 |
+
"eval_samples_per_second": 90.112,
|
| 559 |
+
"eval_steps_per_second": 2.823,
|
| 560 |
+
"step": 800
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 0.74,
|
| 564 |
+
"learning_rate": 9.276756571035904e-05,
|
| 565 |
+
"loss": 632.5693,
|
| 566 |
+
"step": 810
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 0.75,
|
| 570 |
+
"learning_rate": 9.39128442993758e-05,
|
| 571 |
+
"loss": 633.8072,
|
| 572 |
+
"step": 820
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.76,
|
| 576 |
+
"learning_rate": 9.50581228883926e-05,
|
| 577 |
+
"loss": 625.9993,
|
| 578 |
+
"step": 830
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.77,
|
| 582 |
+
"learning_rate": 9.620340147740937e-05,
|
| 583 |
+
"loss": 616.1023,
|
| 584 |
+
"step": 840
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.78,
|
| 588 |
+
"learning_rate": 9.734868006642615e-05,
|
| 589 |
+
"loss": 542.3134,
|
| 590 |
+
"step": 850
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.79,
|
| 594 |
+
"learning_rate": 9.849395865544293e-05,
|
| 595 |
+
"loss": 510.9979,
|
| 596 |
+
"step": 860
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 0.8,
|
| 600 |
+
"learning_rate": 9.96392372444597e-05,
|
| 601 |
+
"loss": 527.1517,
|
| 602 |
+
"step": 870
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.81,
|
| 606 |
+
"learning_rate": 0.00010078451583347648,
|
| 607 |
+
"loss": 595.9219,
|
| 608 |
+
"step": 880
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 0.82,
|
| 612 |
+
"learning_rate": 0.00010192979442249326,
|
| 613 |
+
"loss": 480.4347,
|
| 614 |
+
"step": 890
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 0.82,
|
| 618 |
+
"learning_rate": 0.00010307507301151004,
|
| 619 |
+
"loss": 487.5384,
|
| 620 |
+
"step": 900
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.82,
|
| 624 |
+
"eval_RMSE": 34.58900833129883,
|
| 625 |
+
"eval_loss": 1558.3662109375,
|
| 626 |
+
"eval_runtime": 89.9793,
|
| 627 |
+
"eval_samples_per_second": 90.81,
|
| 628 |
+
"eval_steps_per_second": 2.845,
|
| 629 |
+
"step": 900
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 0.83,
|
| 633 |
+
"learning_rate": 0.00010422035160052684,
|
| 634 |
+
"loss": 538.8044,
|
| 635 |
+
"step": 910
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 0.84,
|
| 639 |
+
"learning_rate": 0.0001053656301895436,
|
| 640 |
+
"loss": 438.5807,
|
| 641 |
+
"step": 920
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.85,
|
| 645 |
+
"learning_rate": 0.00010651090877856037,
|
| 646 |
+
"loss": 492.5742,
|
| 647 |
+
"step": 930
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.86,
|
| 651 |
+
"learning_rate": 0.00010765618736757717,
|
| 652 |
+
"loss": 368.3077,
|
| 653 |
+
"step": 940
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.87,
|
| 657 |
+
"learning_rate": 0.00010880146595659392,
|
| 658 |
+
"loss": 374.0631,
|
| 659 |
+
"step": 950
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 0.88,
|
| 663 |
+
"learning_rate": 0.00010994674454561072,
|
| 664 |
+
"loss": 505.5752,
|
| 665 |
+
"step": 960
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 0.89,
|
| 669 |
+
"learning_rate": 0.0001110920231346275,
|
| 670 |
+
"loss": 392.0846,
|
| 671 |
+
"step": 970
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 0.9,
|
| 675 |
+
"learning_rate": 0.00011223730172364425,
|
| 676 |
+
"loss": 348.8105,
|
| 677 |
+
"step": 980
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 0.91,
|
| 681 |
+
"learning_rate": 0.00011338258031266105,
|
| 682 |
+
"loss": 351.132,
|
| 683 |
+
"step": 990
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 0.92,
|
| 687 |
+
"learning_rate": 0.00011452785890167783,
|
| 688 |
+
"loss": 363.1367,
|
| 689 |
+
"step": 1000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.92,
|
| 693 |
+
"eval_RMSE": 31.56056785583496,
|
| 694 |
+
"eval_loss": 1274.2431640625,
|
| 695 |
+
"eval_runtime": 89.8821,
|
| 696 |
+
"eval_samples_per_second": 90.908,
|
| 697 |
+
"eval_steps_per_second": 2.848,
|
| 698 |
+
"step": 1000
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 0.92,
|
| 702 |
+
"learning_rate": 0.0001156731374906946,
|
| 703 |
+
"loss": 296.5691,
|
| 704 |
+
"step": 1010
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 0.93,
|
| 708 |
+
"learning_rate": 0.00011681841607971138,
|
| 709 |
+
"loss": 357.6329,
|
| 710 |
+
"step": 1020
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.94,
|
| 714 |
+
"learning_rate": 0.00011796369466872816,
|
| 715 |
+
"loss": 361.3535,
|
| 716 |
+
"step": 1030
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.95,
|
| 720 |
+
"learning_rate": 0.00011910897325774494,
|
| 721 |
+
"loss": 322.6138,
|
| 722 |
+
"step": 1040
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.96,
|
| 726 |
+
"learning_rate": 0.00012025425184676171,
|
| 727 |
+
"loss": 405.8568,
|
| 728 |
+
"step": 1050
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 0.97,
|
| 732 |
+
"learning_rate": 0.00012139953043577849,
|
| 733 |
+
"loss": 352.4393,
|
| 734 |
+
"step": 1060
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 0.98,
|
| 738 |
+
"learning_rate": 0.00012254480902479527,
|
| 739 |
+
"loss": 367.3307,
|
| 740 |
+
"step": 1070
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 0.99,
|
| 744 |
+
"learning_rate": 0.00012369008761381204,
|
| 745 |
+
"loss": 300.1264,
|
| 746 |
+
"step": 1080
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 1.0,
|
| 750 |
+
"learning_rate": 0.00012483536620282884,
|
| 751 |
+
"loss": 344.8634,
|
| 752 |
+
"step": 1090
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 1.01,
|
| 756 |
+
"learning_rate": 0.0001259806447918456,
|
| 757 |
+
"loss": 275.3597,
|
| 758 |
+
"step": 1100
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.01,
|
| 762 |
+
"eval_RMSE": 28.991735458374023,
|
| 763 |
+
"eval_loss": 1044.326416015625,
|
| 764 |
+
"eval_runtime": 90.499,
|
| 765 |
+
"eval_samples_per_second": 90.288,
|
| 766 |
+
"eval_steps_per_second": 2.829,
|
| 767 |
+
"step": 1100
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 1.02,
|
| 771 |
+
"learning_rate": 0.00012712592338086237,
|
| 772 |
+
"loss": 354.2523,
|
| 773 |
+
"step": 1110
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.03,
|
| 777 |
+
"learning_rate": 0.00012827120196987917,
|
| 778 |
+
"loss": 283.569,
|
| 779 |
+
"step": 1120
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 1.03,
|
| 783 |
+
"learning_rate": 0.00012941648055889594,
|
| 784 |
+
"loss": 248.2783,
|
| 785 |
+
"step": 1130
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 1.04,
|
| 789 |
+
"learning_rate": 0.00013056175914791273,
|
| 790 |
+
"loss": 262.5986,
|
| 791 |
+
"step": 1140
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 1.05,
|
| 795 |
+
"learning_rate": 0.0001317070377369295,
|
| 796 |
+
"loss": 294.735,
|
| 797 |
+
"step": 1150
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 1.06,
|
| 801 |
+
"learning_rate": 0.0001328523163259463,
|
| 802 |
+
"loss": 301.2489,
|
| 803 |
+
"step": 1160
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 1.07,
|
| 807 |
+
"learning_rate": 0.00013399759491496306,
|
| 808 |
+
"loss": 252.3795,
|
| 809 |
+
"step": 1170
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.08,
|
| 813 |
+
"learning_rate": 0.00013514287350397983,
|
| 814 |
+
"loss": 247.8596,
|
| 815 |
+
"step": 1180
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 1.09,
|
| 819 |
+
"learning_rate": 0.00013628815209299662,
|
| 820 |
+
"loss": 253.8303,
|
| 821 |
+
"step": 1190
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 1.1,
|
| 825 |
+
"learning_rate": 0.0001374334306820134,
|
| 826 |
+
"loss": 251.5496,
|
| 827 |
+
"step": 1200
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 1.1,
|
| 831 |
+
"eval_RMSE": 27.961669921875,
|
| 832 |
+
"eval_loss": 858.2109985351562,
|
| 833 |
+
"eval_runtime": 90.7351,
|
| 834 |
+
"eval_samples_per_second": 90.053,
|
| 835 |
+
"eval_steps_per_second": 2.821,
|
| 836 |
+
"step": 1200
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 1.11,
|
| 840 |
+
"learning_rate": 0.00013857870927103016,
|
| 841 |
+
"loss": 207.0559,
|
| 842 |
+
"step": 1210
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 1.12,
|
| 846 |
+
"learning_rate": 0.00013972398786004695,
|
| 847 |
+
"loss": 231.4458,
|
| 848 |
+
"step": 1220
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 1.13,
|
| 852 |
+
"learning_rate": 0.00014086926644906372,
|
| 853 |
+
"loss": 252.8603,
|
| 854 |
+
"step": 1230
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 1.14,
|
| 858 |
+
"learning_rate": 0.00014201454503808052,
|
| 859 |
+
"loss": 204.8269,
|
| 860 |
+
"step": 1240
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.14,
|
| 864 |
+
"learning_rate": 0.00014315982362709729,
|
| 865 |
+
"loss": 230.3754,
|
| 866 |
+
"step": 1250
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.15,
|
| 870 |
+
"learning_rate": 0.00014430510221611405,
|
| 871 |
+
"loss": 170.1902,
|
| 872 |
+
"step": 1260
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 1.16,
|
| 876 |
+
"learning_rate": 0.00014545038080513085,
|
| 877 |
+
"loss": 207.8812,
|
| 878 |
+
"step": 1270
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 1.17,
|
| 882 |
+
"learning_rate": 0.00014659565939414762,
|
| 883 |
+
"loss": 250.5713,
|
| 884 |
+
"step": 1280
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 1.18,
|
| 888 |
+
"learning_rate": 0.00014774093798316438,
|
| 889 |
+
"loss": 210.8263,
|
| 890 |
+
"step": 1290
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.19,
|
| 894 |
+
"learning_rate": 0.00014888621657218118,
|
| 895 |
+
"loss": 232.956,
|
| 896 |
+
"step": 1300
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 1.19,
|
| 900 |
+
"eval_RMSE": 28.184946060180664,
|
| 901 |
+
"eval_loss": 733.7390747070312,
|
| 902 |
+
"eval_runtime": 89.6232,
|
| 903 |
+
"eval_samples_per_second": 91.171,
|
| 904 |
+
"eval_steps_per_second": 2.856,
|
| 905 |
+
"step": 1300
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 1.2,
|
| 909 |
+
"learning_rate": 0.00015003149516119795,
|
| 910 |
+
"loss": 209.1999,
|
| 911 |
+
"step": 1310
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 1.21,
|
| 915 |
+
"learning_rate": 0.00015117677375021474,
|
| 916 |
+
"loss": 219.4115,
|
| 917 |
+
"step": 1320
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 1.22,
|
| 921 |
+
"learning_rate": 0.0001523220523392315,
|
| 922 |
+
"loss": 142.051,
|
| 923 |
+
"step": 1330
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 1.23,
|
| 927 |
+
"learning_rate": 0.0001534673309282483,
|
| 928 |
+
"loss": 182.2509,
|
| 929 |
+
"step": 1340
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 1.24,
|
| 933 |
+
"learning_rate": 0.00015461260951726507,
|
| 934 |
+
"loss": 168.6032,
|
| 935 |
+
"step": 1350
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 1.25,
|
| 939 |
+
"learning_rate": 0.00015575788810628184,
|
| 940 |
+
"loss": 171.8442,
|
| 941 |
+
"step": 1360
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 1.25,
|
| 945 |
+
"learning_rate": 0.00015690316669529864,
|
| 946 |
+
"loss": 171.8282,
|
| 947 |
+
"step": 1370
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 1.26,
|
| 951 |
+
"learning_rate": 0.0001580484452843154,
|
| 952 |
+
"loss": 186.681,
|
| 953 |
+
"step": 1380
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 1.27,
|
| 957 |
+
"learning_rate": 0.00015919372387333217,
|
| 958 |
+
"loss": 173.0698,
|
| 959 |
+
"step": 1390
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 1.28,
|
| 963 |
+
"learning_rate": 0.00016033900246234897,
|
| 964 |
+
"loss": 148.5199,
|
| 965 |
+
"step": 1400
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 1.28,
|
| 969 |
+
"eval_RMSE": 27.53377914428711,
|
| 970 |
+
"eval_loss": 608.1200561523438,
|
| 971 |
+
"eval_runtime": 87.4424,
|
| 972 |
+
"eval_samples_per_second": 93.444,
|
| 973 |
+
"eval_steps_per_second": 2.928,
|
| 974 |
+
"step": 1400
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 1.29,
|
| 978 |
+
"learning_rate": 0.00016148428105136573,
|
| 979 |
+
"loss": 195.879,
|
| 980 |
+
"step": 1410
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 1.3,
|
| 984 |
+
"learning_rate": 0.0001626295596403825,
|
| 985 |
+
"loss": 161.7053,
|
| 986 |
+
"step": 1420
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 1.31,
|
| 990 |
+
"learning_rate": 0.0001637748382293993,
|
| 991 |
+
"loss": 144.2296,
|
| 992 |
+
"step": 1430
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 1.32,
|
| 996 |
+
"learning_rate": 0.00016492011681841607,
|
| 997 |
+
"loss": 146.4364,
|
| 998 |
+
"step": 1440
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 1.33,
|
| 1002 |
+
"learning_rate": 0.00016606539540743283,
|
| 1003 |
+
"loss": 144.3109,
|
| 1004 |
+
"step": 1450
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 1.34,
|
| 1008 |
+
"learning_rate": 0.00016721067399644963,
|
| 1009 |
+
"loss": 184.0622,
|
| 1010 |
+
"step": 1460
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 1.35,
|
| 1014 |
+
"learning_rate": 0.00016835595258546642,
|
| 1015 |
+
"loss": 127.7919,
|
| 1016 |
+
"step": 1470
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 1.36,
|
| 1020 |
+
"learning_rate": 0.00016950123117448316,
|
| 1021 |
+
"loss": 133.7518,
|
| 1022 |
+
"step": 1480
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 1.36,
|
| 1026 |
+
"learning_rate": 0.00017064650976349996,
|
| 1027 |
+
"loss": 112.2735,
|
| 1028 |
+
"step": 1490
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 1.37,
|
| 1032 |
+
"learning_rate": 0.00017179178835251675,
|
| 1033 |
+
"loss": 126.9139,
|
| 1034 |
+
"step": 1500
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 1.37,
|
| 1038 |
+
"eval_RMSE": 26.30722999572754,
|
| 1039 |
+
"eval_loss": 453.3481750488281,
|
| 1040 |
+
"eval_runtime": 88.0738,
|
| 1041 |
+
"eval_samples_per_second": 92.774,
|
| 1042 |
+
"eval_steps_per_second": 2.907,
|
| 1043 |
+
"step": 1500
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 1.38,
|
| 1047 |
+
"learning_rate": 0.00017293706694153352,
|
| 1048 |
+
"loss": 134.5855,
|
| 1049 |
+
"step": 1510
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 1.39,
|
| 1053 |
+
"learning_rate": 0.0001740823455305503,
|
| 1054 |
+
"loss": 124.9829,
|
| 1055 |
+
"step": 1520
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 1.4,
|
| 1059 |
+
"learning_rate": 0.00017522762411956708,
|
| 1060 |
+
"loss": 111.1943,
|
| 1061 |
+
"step": 1530
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 1.41,
|
| 1065 |
+
"learning_rate": 0.00017637290270858385,
|
| 1066 |
+
"loss": 124.6046,
|
| 1067 |
+
"step": 1540
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 1.42,
|
| 1071 |
+
"learning_rate": 0.00017751818129760062,
|
| 1072 |
+
"loss": 97.5316,
|
| 1073 |
+
"step": 1550
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 1.43,
|
| 1077 |
+
"learning_rate": 0.00017866345988661742,
|
| 1078 |
+
"loss": 130.0623,
|
| 1079 |
+
"step": 1560
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 1.44,
|
| 1083 |
+
"learning_rate": 0.0001798087384756342,
|
| 1084 |
+
"loss": 140.2047,
|
| 1085 |
+
"step": 1570
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 1.45,
|
| 1089 |
+
"learning_rate": 0.00018095401706465095,
|
| 1090 |
+
"loss": 111.0409,
|
| 1091 |
+
"step": 1580
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 1.46,
|
| 1095 |
+
"learning_rate": 0.00018209929565366775,
|
| 1096 |
+
"loss": 112.2171,
|
| 1097 |
+
"step": 1590
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 1.47,
|
| 1101 |
+
"learning_rate": 0.00018324457424268454,
|
| 1102 |
+
"loss": 113.2355,
|
| 1103 |
+
"step": 1600
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 1.47,
|
| 1107 |
+
"eval_RMSE": 26.593936920166016,
|
| 1108 |
+
"eval_loss": 397.5113220214844,
|
| 1109 |
+
"eval_runtime": 89.5143,
|
| 1110 |
+
"eval_samples_per_second": 91.282,
|
| 1111 |
+
"eval_steps_per_second": 2.86,
|
| 1112 |
+
"step": 1600
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 1.47,
|
| 1116 |
+
"learning_rate": 0.00018438985283170128,
|
| 1117 |
+
"loss": 121.8388,
|
| 1118 |
+
"step": 1610
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 1.48,
|
| 1122 |
+
"learning_rate": 0.00018553513142071808,
|
| 1123 |
+
"loss": 116.3681,
|
| 1124 |
+
"step": 1620
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 1.49,
|
| 1128 |
+
"learning_rate": 0.00018668041000973487,
|
| 1129 |
+
"loss": 113.7981,
|
| 1130 |
+
"step": 1630
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 1.5,
|
| 1134 |
+
"learning_rate": 0.0001878256885987516,
|
| 1135 |
+
"loss": 119.2628,
|
| 1136 |
+
"step": 1640
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 1.51,
|
| 1140 |
+
"learning_rate": 0.0001889709671877684,
|
| 1141 |
+
"loss": 93.8756,
|
| 1142 |
+
"step": 1650
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 1.52,
|
| 1146 |
+
"learning_rate": 0.0001901162457767852,
|
| 1147 |
+
"loss": 118.5608,
|
| 1148 |
+
"step": 1660
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 1.53,
|
| 1152 |
+
"learning_rate": 0.00019126152436580197,
|
| 1153 |
+
"loss": 101.4624,
|
| 1154 |
+
"step": 1670
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 1.54,
|
| 1158 |
+
"learning_rate": 0.00019240680295481874,
|
| 1159 |
+
"loss": 96.8405,
|
| 1160 |
+
"step": 1680
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 1.55,
|
| 1164 |
+
"learning_rate": 0.00019355208154383553,
|
| 1165 |
+
"loss": 93.3038,
|
| 1166 |
+
"step": 1690
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 1.56,
|
| 1170 |
+
"learning_rate": 0.0001946973601328523,
|
| 1171 |
+
"loss": 99.5798,
|
| 1172 |
+
"step": 1700
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 1.56,
|
| 1176 |
+
"eval_RMSE": 25.650577545166016,
|
| 1177 |
+
"eval_loss": 304.5827331542969,
|
| 1178 |
+
"eval_runtime": 88.1891,
|
| 1179 |
+
"eval_samples_per_second": 92.653,
|
| 1180 |
+
"eval_steps_per_second": 2.903,
|
| 1181 |
+
"step": 1700
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 1.57,
|
| 1185 |
+
"learning_rate": 0.00019584263872186907,
|
| 1186 |
+
"loss": 116.748,
|
| 1187 |
+
"step": 1710
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 1.58,
|
| 1191 |
+
"learning_rate": 0.00019698791731088586,
|
| 1192 |
+
"loss": 87.1141,
|
| 1193 |
+
"step": 1720
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 1.58,
|
| 1197 |
+
"learning_rate": 0.00019813319589990266,
|
| 1198 |
+
"loss": 97.8321,
|
| 1199 |
+
"step": 1730
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 1.59,
|
| 1203 |
+
"learning_rate": 0.0001992784744889194,
|
| 1204 |
+
"loss": 79.7827,
|
| 1205 |
+
"step": 1740
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 1.6,
|
| 1209 |
+
"learning_rate": 0.00019995291632467376,
|
| 1210 |
+
"loss": 98.2981,
|
| 1211 |
+
"step": 1750
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 1.61,
|
| 1215 |
+
"learning_rate": 0.00019982566314811633,
|
| 1216 |
+
"loss": 68.1837,
|
| 1217 |
+
"step": 1760
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 1.62,
|
| 1221 |
+
"learning_rate": 0.0001996984099715589,
|
| 1222 |
+
"loss": 88.1333,
|
| 1223 |
+
"step": 1770
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 1.63,
|
| 1227 |
+
"learning_rate": 0.0001995711567950015,
|
| 1228 |
+
"loss": 84.4777,
|
| 1229 |
+
"step": 1780
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 1.64,
|
| 1233 |
+
"learning_rate": 0.00019944390361844406,
|
| 1234 |
+
"loss": 75.6494,
|
| 1235 |
+
"step": 1790
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 1.65,
|
| 1239 |
+
"learning_rate": 0.00019931665044188666,
|
| 1240 |
+
"loss": 101.6581,
|
| 1241 |
+
"step": 1800
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 1.65,
|
| 1245 |
+
"eval_RMSE": 24.347003936767578,
|
| 1246 |
+
"eval_loss": 246.870849609375,
|
| 1247 |
+
"eval_runtime": 91.7415,
|
| 1248 |
+
"eval_samples_per_second": 89.066,
|
| 1249 |
+
"eval_steps_per_second": 2.79,
|
| 1250 |
+
"step": 1800
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 1.66,
|
| 1254 |
+
"learning_rate": 0.00019918939726532923,
|
| 1255 |
+
"loss": 79.9642,
|
| 1256 |
+
"step": 1810
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 1.67,
|
| 1260 |
+
"learning_rate": 0.00019906214408877182,
|
| 1261 |
+
"loss": 67.1507,
|
| 1262 |
+
"step": 1820
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 1.68,
|
| 1266 |
+
"learning_rate": 0.0001989348909122144,
|
| 1267 |
+
"loss": 78.1831,
|
| 1268 |
+
"step": 1830
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 1.68,
|
| 1272 |
+
"learning_rate": 0.000198807637735657,
|
| 1273 |
+
"loss": 72.3693,
|
| 1274 |
+
"step": 1840
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 1.69,
|
| 1278 |
+
"learning_rate": 0.00019868038455909956,
|
| 1279 |
+
"loss": 73.2772,
|
| 1280 |
+
"step": 1850
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 1.7,
|
| 1284 |
+
"learning_rate": 0.00019855313138254216,
|
| 1285 |
+
"loss": 73.4863,
|
| 1286 |
+
"step": 1860
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 1.71,
|
| 1290 |
+
"learning_rate": 0.00019842587820598472,
|
| 1291 |
+
"loss": 68.861,
|
| 1292 |
+
"step": 1870
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 1.72,
|
| 1296 |
+
"learning_rate": 0.0001982986250294273,
|
| 1297 |
+
"loss": 70.2734,
|
| 1298 |
+
"step": 1880
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 1.73,
|
| 1302 |
+
"learning_rate": 0.00019817137185286986,
|
| 1303 |
+
"loss": 80.5998,
|
| 1304 |
+
"step": 1890
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 1.74,
|
| 1308 |
+
"learning_rate": 0.00019804411867631246,
|
| 1309 |
+
"loss": 74.3472,
|
| 1310 |
+
"step": 1900
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 1.74,
|
| 1314 |
+
"eval_RMSE": 25.971708297729492,
|
| 1315 |
+
"eval_loss": 242.7896270751953,
|
| 1316 |
+
"eval_runtime": 89.3685,
|
| 1317 |
+
"eval_samples_per_second": 91.43,
|
| 1318 |
+
"eval_steps_per_second": 2.865,
|
| 1319 |
+
"step": 1900
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 1.75,
|
| 1323 |
+
"learning_rate": 0.00019791686549975503,
|
| 1324 |
+
"loss": 88.6437,
|
| 1325 |
+
"step": 1910
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 1.76,
|
| 1329 |
+
"learning_rate": 0.00019778961232319763,
|
| 1330 |
+
"loss": 95.2708,
|
| 1331 |
+
"step": 1920
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 1.77,
|
| 1335 |
+
"learning_rate": 0.0001976623591466402,
|
| 1336 |
+
"loss": 114.3341,
|
| 1337 |
+
"step": 1930
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 1.78,
|
| 1341 |
+
"learning_rate": 0.0001975351059700828,
|
| 1342 |
+
"loss": 67.5784,
|
| 1343 |
+
"step": 1940
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 1.79,
|
| 1347 |
+
"learning_rate": 0.00019740785279352536,
|
| 1348 |
+
"loss": 73.2939,
|
| 1349 |
+
"step": 1950
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 1.79,
|
| 1353 |
+
"learning_rate": 0.00019728059961696793,
|
| 1354 |
+
"loss": 68.14,
|
| 1355 |
+
"step": 1960
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 1.8,
|
| 1359 |
+
"learning_rate": 0.00019715334644041053,
|
| 1360 |
+
"loss": 69.5453,
|
| 1361 |
+
"step": 1970
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 1.81,
|
| 1365 |
+
"learning_rate": 0.0001970260932638531,
|
| 1366 |
+
"loss": 81.5987,
|
| 1367 |
+
"step": 1980
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 1.82,
|
| 1371 |
+
"learning_rate": 0.0001968988400872957,
|
| 1372 |
+
"loss": 71.576,
|
| 1373 |
+
"step": 1990
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 1.83,
|
| 1377 |
+
"learning_rate": 0.00019677158691073826,
|
| 1378 |
+
"loss": 77.3393,
|
| 1379 |
+
"step": 2000
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 1.83,
|
| 1383 |
+
"eval_RMSE": 25.02290153503418,
|
| 1384 |
+
"eval_loss": 158.2856903076172,
|
| 1385 |
+
"eval_runtime": 91.421,
|
| 1386 |
+
"eval_samples_per_second": 89.378,
|
| 1387 |
+
"eval_steps_per_second": 2.8,
|
| 1388 |
+
"step": 2000
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 1.84,
|
| 1392 |
+
"learning_rate": 0.00019664433373418083,
|
| 1393 |
+
"loss": 71.151,
|
| 1394 |
+
"step": 2010
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 1.85,
|
| 1398 |
+
"learning_rate": 0.0001965170805576234,
|
| 1399 |
+
"loss": 57.0214,
|
| 1400 |
+
"step": 2020
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 1.86,
|
| 1404 |
+
"learning_rate": 0.000196389827381066,
|
| 1405 |
+
"loss": 60.4723,
|
| 1406 |
+
"step": 2030
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 1.87,
|
| 1410 |
+
"learning_rate": 0.00019626257420450857,
|
| 1411 |
+
"loss": 61.1471,
|
| 1412 |
+
"step": 2040
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 1.88,
|
| 1416 |
+
"learning_rate": 0.00019613532102795116,
|
| 1417 |
+
"loss": 64.6843,
|
| 1418 |
+
"step": 2050
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 1.89,
|
| 1422 |
+
"learning_rate": 0.00019600806785139373,
|
| 1423 |
+
"loss": 80.3143,
|
| 1424 |
+
"step": 2060
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 1.9,
|
| 1428 |
+
"learning_rate": 0.00019588081467483633,
|
| 1429 |
+
"loss": 61.2826,
|
| 1430 |
+
"step": 2070
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 1.9,
|
| 1434 |
+
"learning_rate": 0.0001957535614982789,
|
| 1435 |
+
"loss": 87.4192,
|
| 1436 |
+
"step": 2080
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 1.91,
|
| 1440 |
+
"learning_rate": 0.0001956263083217215,
|
| 1441 |
+
"loss": 80.7123,
|
| 1442 |
+
"step": 2090
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 1.92,
|
| 1446 |
+
"learning_rate": 0.00019549905514516406,
|
| 1447 |
+
"loss": 69.8205,
|
| 1448 |
+
"step": 2100
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 1.92,
|
| 1452 |
+
"eval_RMSE": 26.108875274658203,
|
| 1453 |
+
"eval_loss": 163.57669067382812,
|
| 1454 |
+
"eval_runtime": 88.3567,
|
| 1455 |
+
"eval_samples_per_second": 92.477,
|
| 1456 |
+
"eval_steps_per_second": 2.897,
|
| 1457 |
+
"step": 2100
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 1.93,
|
| 1461 |
+
"learning_rate": 0.00019537180196860666,
|
| 1462 |
+
"loss": 68.615,
|
| 1463 |
+
"step": 2110
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 1.94,
|
| 1467 |
+
"learning_rate": 0.00019524454879204923,
|
| 1468 |
+
"loss": 66.9629,
|
| 1469 |
+
"step": 2120
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 1.95,
|
| 1473 |
+
"learning_rate": 0.00019511729561549182,
|
| 1474 |
+
"loss": 66.9707,
|
| 1475 |
+
"step": 2130
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 1.96,
|
| 1479 |
+
"learning_rate": 0.00019499004243893437,
|
| 1480 |
+
"loss": 61.8375,
|
| 1481 |
+
"step": 2140
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 1.97,
|
| 1485 |
+
"learning_rate": 0.00019486278926237696,
|
| 1486 |
+
"loss": 86.5031,
|
| 1487 |
+
"step": 2150
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 1.98,
|
| 1491 |
+
"learning_rate": 0.00019473553608581953,
|
| 1492 |
+
"loss": 59.317,
|
| 1493 |
+
"step": 2160
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 1.99,
|
| 1497 |
+
"learning_rate": 0.00019460828290926213,
|
| 1498 |
+
"loss": 46.4391,
|
| 1499 |
+
"step": 2170
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 2.0,
|
| 1503 |
+
"learning_rate": 0.0001944810297327047,
|
| 1504 |
+
"loss": 66.9274,
|
| 1505 |
+
"step": 2180
|
| 1506 |
+
},
|
| 1507 |
+
{
|
| 1508 |
+
"epoch": 2.01,
|
| 1509 |
+
"learning_rate": 0.0001943537765561473,
|
| 1510 |
+
"loss": 60.0429,
|
| 1511 |
+
"step": 2190
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 2.01,
|
| 1515 |
+
"learning_rate": 0.00019422652337958986,
|
| 1516 |
+
"loss": 49.6239,
|
| 1517 |
+
"step": 2200
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 2.01,
|
| 1521 |
+
"eval_RMSE": 25.317358016967773,
|
| 1522 |
+
"eval_loss": 155.91978454589844,
|
| 1523 |
+
"eval_runtime": 89.937,
|
| 1524 |
+
"eval_samples_per_second": 90.852,
|
| 1525 |
+
"eval_steps_per_second": 2.846,
|
| 1526 |
+
"step": 2200
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 2.02,
|
| 1530 |
+
"learning_rate": 0.00019409927020303243,
|
| 1531 |
+
"loss": 48.8787,
|
| 1532 |
+
"step": 2210
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 2.03,
|
| 1536 |
+
"learning_rate": 0.00019397201702647503,
|
| 1537 |
+
"loss": 54.4583,
|
| 1538 |
+
"step": 2220
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 2.04,
|
| 1542 |
+
"learning_rate": 0.0001938447638499176,
|
| 1543 |
+
"loss": 51.7232,
|
| 1544 |
+
"step": 2230
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 2.05,
|
| 1548 |
+
"learning_rate": 0.0001937175106733602,
|
| 1549 |
+
"loss": 55.3078,
|
| 1550 |
+
"step": 2240
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 2.06,
|
| 1554 |
+
"learning_rate": 0.00019359025749680276,
|
| 1555 |
+
"loss": 52.4878,
|
| 1556 |
+
"step": 2250
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 2.07,
|
| 1560 |
+
"learning_rate": 0.00019346300432024533,
|
| 1561 |
+
"loss": 51.498,
|
| 1562 |
+
"step": 2260
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 2.08,
|
| 1566 |
+
"learning_rate": 0.0001933357511436879,
|
| 1567 |
+
"loss": 55.693,
|
| 1568 |
+
"step": 2270
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 2.09,
|
| 1572 |
+
"learning_rate": 0.0001932084979671305,
|
| 1573 |
+
"loss": 60.7052,
|
| 1574 |
+
"step": 2280
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 2.1,
|
| 1578 |
+
"learning_rate": 0.00019308124479057307,
|
| 1579 |
+
"loss": 57.5329,
|
| 1580 |
+
"step": 2290
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 2.11,
|
| 1584 |
+
"learning_rate": 0.00019295399161401567,
|
| 1585 |
+
"loss": 49.7357,
|
| 1586 |
+
"step": 2300
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 2.11,
|
| 1590 |
+
"eval_RMSE": 25.681665420532227,
|
| 1591 |
+
"eval_loss": 113.43367004394531,
|
| 1592 |
+
"eval_runtime": 88.8092,
|
| 1593 |
+
"eval_samples_per_second": 92.006,
|
| 1594 |
+
"eval_steps_per_second": 2.883,
|
| 1595 |
+
"step": 2300
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 2.12,
|
| 1599 |
+
"learning_rate": 0.00019282673843745823,
|
| 1600 |
+
"loss": 40.8768,
|
| 1601 |
+
"step": 2310
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 2.12,
|
| 1605 |
+
"learning_rate": 0.00019269948526090083,
|
| 1606 |
+
"loss": 51.1638,
|
| 1607 |
+
"step": 2320
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 2.13,
|
| 1611 |
+
"learning_rate": 0.0001925722320843434,
|
| 1612 |
+
"loss": 53.8347,
|
| 1613 |
+
"step": 2330
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 2.14,
|
| 1617 |
+
"learning_rate": 0.000192444978907786,
|
| 1618 |
+
"loss": 61.9818,
|
| 1619 |
+
"step": 2340
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 2.15,
|
| 1623 |
+
"learning_rate": 0.00019231772573122857,
|
| 1624 |
+
"loss": 65.0766,
|
| 1625 |
+
"step": 2350
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 2.16,
|
| 1629 |
+
"learning_rate": 0.00019219047255467116,
|
| 1630 |
+
"loss": 54.4025,
|
| 1631 |
+
"step": 2360
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 2.17,
|
| 1635 |
+
"learning_rate": 0.00019206321937811373,
|
| 1636 |
+
"loss": 43.9705,
|
| 1637 |
+
"step": 2370
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 2.18,
|
| 1641 |
+
"learning_rate": 0.00019193596620155633,
|
| 1642 |
+
"loss": 58.1021,
|
| 1643 |
+
"step": 2380
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 2.19,
|
| 1647 |
+
"learning_rate": 0.00019180871302499887,
|
| 1648 |
+
"loss": 41.1698,
|
| 1649 |
+
"step": 2390
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 2.2,
|
| 1653 |
+
"learning_rate": 0.00019168145984844147,
|
| 1654 |
+
"loss": 60.9505,
|
| 1655 |
+
"step": 2400
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 2.2,
|
| 1659 |
+
"eval_RMSE": 25.213809967041016,
|
| 1660 |
+
"eval_loss": 128.33050537109375,
|
| 1661 |
+
"eval_runtime": 89.5513,
|
| 1662 |
+
"eval_samples_per_second": 91.244,
|
| 1663 |
+
"eval_steps_per_second": 2.859,
|
| 1664 |
+
"step": 2400
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 2.21,
|
| 1668 |
+
"learning_rate": 0.00019155420667188404,
|
| 1669 |
+
"loss": 50.3419,
|
| 1670 |
+
"step": 2410
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 2.22,
|
| 1674 |
+
"learning_rate": 0.00019142695349532663,
|
| 1675 |
+
"loss": 34.2566,
|
| 1676 |
+
"step": 2420
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 2.23,
|
| 1680 |
+
"learning_rate": 0.0001912997003187692,
|
| 1681 |
+
"loss": 45.824,
|
| 1682 |
+
"step": 2430
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 2.23,
|
| 1686 |
+
"learning_rate": 0.0001911724471422118,
|
| 1687 |
+
"loss": 51.5597,
|
| 1688 |
+
"step": 2440
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 2.24,
|
| 1692 |
+
"learning_rate": 0.00019104519396565437,
|
| 1693 |
+
"loss": 63.5468,
|
| 1694 |
+
"step": 2450
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 2.25,
|
| 1698 |
+
"learning_rate": 0.00019091794078909694,
|
| 1699 |
+
"loss": 47.8715,
|
| 1700 |
+
"step": 2460
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 2.26,
|
| 1704 |
+
"learning_rate": 0.00019079068761253953,
|
| 1705 |
+
"loss": 56.6548,
|
| 1706 |
+
"step": 2470
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 2.27,
|
| 1710 |
+
"learning_rate": 0.0001906634344359821,
|
| 1711 |
+
"loss": 46.0047,
|
| 1712 |
+
"step": 2480
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 2.28,
|
| 1716 |
+
"learning_rate": 0.0001905361812594247,
|
| 1717 |
+
"loss": 55.3212,
|
| 1718 |
+
"step": 2490
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 2.29,
|
| 1722 |
+
"learning_rate": 0.00019040892808286727,
|
| 1723 |
+
"loss": 49.0666,
|
| 1724 |
+
"step": 2500
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 2.29,
|
| 1728 |
+
"eval_RMSE": 26.342649459838867,
|
| 1729 |
+
"eval_loss": 108.29063415527344,
|
| 1730 |
+
"eval_runtime": 89.6891,
|
| 1731 |
+
"eval_samples_per_second": 91.104,
|
| 1732 |
+
"eval_steps_per_second": 2.854,
|
| 1733 |
+
"step": 2500
|
| 1734 |
+
}
|
| 1735 |
+
],
|
| 1736 |
+
"max_steps": 4368,
|
| 1737 |
+
"num_train_epochs": 4,
|
| 1738 |
+
"total_flos": 9.294770879473766e+18,
|
| 1739 |
+
"trial_name": null,
|
| 1740 |
+
"trial_params": null
|
| 1741 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cc889b6bb588ff492d2a1e703fb0d7fa5bb722178d7b55fc1f24f5814763aadb
|
| 3 |
+
size 3899
|