Commit ·
8ab08ee
1
Parent(s): 39b7ccf
Upload 8 files
Browse files- config.json +52 -0
- optimizer.pt +3 -0
- preprocessor_config.json +24 -0
- pytorch_model.bin +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +1003 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,52 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "microsoft/table-transformer-detection",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"activation_function": "relu",
|
| 5 |
+
"architectures": [
|
| 6 |
+
"TableTransformerForObjectDetection"
|
| 7 |
+
],
|
| 8 |
+
"attention_dropout": 0.0,
|
| 9 |
+
"auxiliary_loss": false,
|
| 10 |
+
"backbone": "resnet18",
|
| 11 |
+
"backbone_config": null,
|
| 12 |
+
"bbox_cost": 5,
|
| 13 |
+
"bbox_loss_coefficient": 5,
|
| 14 |
+
"ce_loss_coefficient": 1,
|
| 15 |
+
"class_cost": 1,
|
| 16 |
+
"d_model": 256,
|
| 17 |
+
"decoder_attention_heads": 8,
|
| 18 |
+
"decoder_ffn_dim": 2048,
|
| 19 |
+
"decoder_layerdrop": 0.0,
|
| 20 |
+
"decoder_layers": 6,
|
| 21 |
+
"dice_loss_coefficient": 1,
|
| 22 |
+
"dilation": false,
|
| 23 |
+
"dropout": 0.1,
|
| 24 |
+
"encoder_attention_heads": 8,
|
| 25 |
+
"encoder_ffn_dim": 2048,
|
| 26 |
+
"encoder_layerdrop": 0.0,
|
| 27 |
+
"encoder_layers": 6,
|
| 28 |
+
"eos_coefficient": 0.4,
|
| 29 |
+
"giou_cost": 2,
|
| 30 |
+
"giou_loss_coefficient": 2,
|
| 31 |
+
"id2label": {
|
| 32 |
+
"0": "table"
|
| 33 |
+
},
|
| 34 |
+
"init_std": 0.02,
|
| 35 |
+
"init_xavier_std": 1.0,
|
| 36 |
+
"is_encoder_decoder": true,
|
| 37 |
+
"label2id": {
|
| 38 |
+
"table": "0"
|
| 39 |
+
},
|
| 40 |
+
"mask_loss_coefficient": 1,
|
| 41 |
+
"max_position_embeddings": 1024,
|
| 42 |
+
"model_type": "table-transformer",
|
| 43 |
+
"num_channels": 3,
|
| 44 |
+
"num_hidden_layers": 6,
|
| 45 |
+
"num_queries": 15,
|
| 46 |
+
"position_embedding_type": "sine",
|
| 47 |
+
"scale_embedding": false,
|
| 48 |
+
"torch_dtype": "float32",
|
| 49 |
+
"transformers_version": "4.33.1",
|
| 50 |
+
"use_pretrained_backbone": true,
|
| 51 |
+
"use_timm_backbone": true
|
| 52 |
+
}
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c79d0a550ce1bf34bb025898b7b4ab13c87503f1a1934f6480167fe948575450
|
| 3 |
+
size 229374213
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"do_pad": true,
|
| 4 |
+
"do_rescale": true,
|
| 5 |
+
"do_resize": true,
|
| 6 |
+
"format": "coco_detection",
|
| 7 |
+
"image_mean": [
|
| 8 |
+
0.485,
|
| 9 |
+
0.456,
|
| 10 |
+
0.406
|
| 11 |
+
],
|
| 12 |
+
"image_processor_type": "DetrFeatureExtractor",
|
| 13 |
+
"image_std": [
|
| 14 |
+
0.229,
|
| 15 |
+
0.224,
|
| 16 |
+
0.225
|
| 17 |
+
],
|
| 18 |
+
"resample": 2,
|
| 19 |
+
"rescale_factor": 0.00392156862745098,
|
| 20 |
+
"size": {
|
| 21 |
+
"longest_edge": 800,
|
| 22 |
+
"shortest_edge": 800
|
| 23 |
+
}
|
| 24 |
+
}
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:52ae1b56f3838651e04e3db35dd132c1859ed1dab8932167d593b16ac0fccb6d
|
| 3 |
+
size 115399113
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4c494dfd420ab22818c3e06a6d9c8af5aa9a80200a3649b19bb0075f06695ac1
|
| 3 |
+
size 14575
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7a4c76bf92d08163654bbfa957013d971483785882393638e6c21a29d7ee4d5d
|
| 3 |
+
size 627
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1003 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1171.4285714285713,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 8200,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 7.14,
|
| 13 |
+
"learning_rate": 9.947619047619048e-05,
|
| 14 |
+
"loss": 0.6717,
|
| 15 |
+
"step": 50
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 14.29,
|
| 19 |
+
"learning_rate": 9.888095238095238e-05,
|
| 20 |
+
"loss": 0.2784,
|
| 21 |
+
"step": 100
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 21.43,
|
| 25 |
+
"learning_rate": 9.828571428571429e-05,
|
| 26 |
+
"loss": 0.145,
|
| 27 |
+
"step": 150
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 28.57,
|
| 31 |
+
"learning_rate": 9.76904761904762e-05,
|
| 32 |
+
"loss": 0.1728,
|
| 33 |
+
"step": 200
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 35.71,
|
| 37 |
+
"learning_rate": 9.70952380952381e-05,
|
| 38 |
+
"loss": 0.1473,
|
| 39 |
+
"step": 250
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 42.86,
|
| 43 |
+
"learning_rate": 9.65e-05,
|
| 44 |
+
"loss": 0.1465,
|
| 45 |
+
"step": 300
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 50.0,
|
| 49 |
+
"learning_rate": 9.59047619047619e-05,
|
| 50 |
+
"loss": 0.1149,
|
| 51 |
+
"step": 350
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 57.14,
|
| 55 |
+
"learning_rate": 9.530952380952381e-05,
|
| 56 |
+
"loss": 0.1282,
|
| 57 |
+
"step": 400
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 64.29,
|
| 61 |
+
"learning_rate": 9.471428571428573e-05,
|
| 62 |
+
"loss": 0.1007,
|
| 63 |
+
"step": 450
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 71.43,
|
| 67 |
+
"learning_rate": 9.411904761904763e-05,
|
| 68 |
+
"loss": 0.1021,
|
| 69 |
+
"step": 500
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 78.57,
|
| 73 |
+
"learning_rate": 9.352380952380952e-05,
|
| 74 |
+
"loss": 0.1168,
|
| 75 |
+
"step": 550
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 85.71,
|
| 79 |
+
"learning_rate": 9.292857142857143e-05,
|
| 80 |
+
"loss": 0.1142,
|
| 81 |
+
"step": 600
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 92.86,
|
| 85 |
+
"learning_rate": 9.233333333333333e-05,
|
| 86 |
+
"loss": 0.1003,
|
| 87 |
+
"step": 650
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 100.0,
|
| 91 |
+
"learning_rate": 9.173809523809525e-05,
|
| 92 |
+
"loss": 0.1018,
|
| 93 |
+
"step": 700
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 107.14,
|
| 97 |
+
"learning_rate": 9.114285714285716e-05,
|
| 98 |
+
"loss": 0.1016,
|
| 99 |
+
"step": 750
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 114.29,
|
| 103 |
+
"learning_rate": 9.054761904761905e-05,
|
| 104 |
+
"loss": 0.0862,
|
| 105 |
+
"step": 800
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 121.43,
|
| 109 |
+
"learning_rate": 8.995238095238095e-05,
|
| 110 |
+
"loss": 0.0949,
|
| 111 |
+
"step": 850
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 128.57,
|
| 115 |
+
"learning_rate": 8.935714285714285e-05,
|
| 116 |
+
"loss": 0.0958,
|
| 117 |
+
"step": 900
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 135.71,
|
| 121 |
+
"learning_rate": 8.876190476190477e-05,
|
| 122 |
+
"loss": 0.0988,
|
| 123 |
+
"step": 950
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 142.86,
|
| 127 |
+
"learning_rate": 8.816666666666668e-05,
|
| 128 |
+
"loss": 0.1019,
|
| 129 |
+
"step": 1000
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 150.0,
|
| 133 |
+
"learning_rate": 8.757142857142857e-05,
|
| 134 |
+
"loss": 0.0828,
|
| 135 |
+
"step": 1050
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 157.14,
|
| 139 |
+
"learning_rate": 8.697619047619047e-05,
|
| 140 |
+
"loss": 0.1038,
|
| 141 |
+
"step": 1100
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 164.29,
|
| 145 |
+
"learning_rate": 8.638095238095239e-05,
|
| 146 |
+
"loss": 0.0786,
|
| 147 |
+
"step": 1150
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 171.43,
|
| 151 |
+
"learning_rate": 8.57857142857143e-05,
|
| 152 |
+
"loss": 0.0899,
|
| 153 |
+
"step": 1200
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 178.57,
|
| 157 |
+
"learning_rate": 8.51904761904762e-05,
|
| 158 |
+
"loss": 0.0902,
|
| 159 |
+
"step": 1250
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 185.71,
|
| 163 |
+
"learning_rate": 8.459523809523809e-05,
|
| 164 |
+
"loss": 0.092,
|
| 165 |
+
"step": 1300
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 192.86,
|
| 169 |
+
"learning_rate": 8.4e-05,
|
| 170 |
+
"loss": 0.1105,
|
| 171 |
+
"step": 1350
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 200.0,
|
| 175 |
+
"learning_rate": 8.340476190476191e-05,
|
| 176 |
+
"loss": 0.1127,
|
| 177 |
+
"step": 1400
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 207.14,
|
| 181 |
+
"learning_rate": 8.280952380952382e-05,
|
| 182 |
+
"loss": 0.0982,
|
| 183 |
+
"step": 1450
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 214.29,
|
| 187 |
+
"learning_rate": 8.221428571428572e-05,
|
| 188 |
+
"loss": 0.0833,
|
| 189 |
+
"step": 1500
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 221.43,
|
| 193 |
+
"learning_rate": 8.161904761904763e-05,
|
| 194 |
+
"loss": 0.0933,
|
| 195 |
+
"step": 1550
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 228.57,
|
| 199 |
+
"learning_rate": 8.102380952380952e-05,
|
| 200 |
+
"loss": 0.0783,
|
| 201 |
+
"step": 1600
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 235.71,
|
| 205 |
+
"learning_rate": 8.042857142857144e-05,
|
| 206 |
+
"loss": 0.0984,
|
| 207 |
+
"step": 1650
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 242.86,
|
| 211 |
+
"learning_rate": 7.983333333333334e-05,
|
| 212 |
+
"loss": 0.0915,
|
| 213 |
+
"step": 1700
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 250.0,
|
| 217 |
+
"learning_rate": 7.923809523809524e-05,
|
| 218 |
+
"loss": 0.0815,
|
| 219 |
+
"step": 1750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 257.14,
|
| 223 |
+
"learning_rate": 7.864285714285715e-05,
|
| 224 |
+
"loss": 0.1008,
|
| 225 |
+
"step": 1800
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 264.29,
|
| 229 |
+
"learning_rate": 7.804761904761905e-05,
|
| 230 |
+
"loss": 0.0739,
|
| 231 |
+
"step": 1850
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 271.43,
|
| 235 |
+
"learning_rate": 7.745238095238096e-05,
|
| 236 |
+
"loss": 0.0777,
|
| 237 |
+
"step": 1900
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 278.57,
|
| 241 |
+
"learning_rate": 7.685714285714286e-05,
|
| 242 |
+
"loss": 0.0946,
|
| 243 |
+
"step": 1950
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 285.71,
|
| 247 |
+
"learning_rate": 7.626190476190477e-05,
|
| 248 |
+
"loss": 0.0688,
|
| 249 |
+
"step": 2000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 292.86,
|
| 253 |
+
"learning_rate": 7.567857142857143e-05,
|
| 254 |
+
"loss": 0.0775,
|
| 255 |
+
"step": 2050
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 300.0,
|
| 259 |
+
"learning_rate": 7.508333333333333e-05,
|
| 260 |
+
"loss": 0.0716,
|
| 261 |
+
"step": 2100
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 307.14,
|
| 265 |
+
"learning_rate": 7.448809523809525e-05,
|
| 266 |
+
"loss": 0.0658,
|
| 267 |
+
"step": 2150
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 314.29,
|
| 271 |
+
"learning_rate": 7.389285714285714e-05,
|
| 272 |
+
"loss": 0.0853,
|
| 273 |
+
"step": 2200
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 321.43,
|
| 277 |
+
"learning_rate": 7.329761904761905e-05,
|
| 278 |
+
"loss": 0.0672,
|
| 279 |
+
"step": 2250
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 328.57,
|
| 283 |
+
"learning_rate": 7.270238095238095e-05,
|
| 284 |
+
"loss": 0.0764,
|
| 285 |
+
"step": 2300
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 335.71,
|
| 289 |
+
"learning_rate": 7.210714285714286e-05,
|
| 290 |
+
"loss": 0.1021,
|
| 291 |
+
"step": 2350
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 342.86,
|
| 295 |
+
"learning_rate": 7.151190476190477e-05,
|
| 296 |
+
"loss": 0.0771,
|
| 297 |
+
"step": 2400
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 350.0,
|
| 301 |
+
"learning_rate": 7.091666666666666e-05,
|
| 302 |
+
"loss": 0.0682,
|
| 303 |
+
"step": 2450
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 357.14,
|
| 307 |
+
"learning_rate": 7.032142857142857e-05,
|
| 308 |
+
"loss": 0.0714,
|
| 309 |
+
"step": 2500
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 364.29,
|
| 313 |
+
"learning_rate": 6.972619047619047e-05,
|
| 314 |
+
"loss": 0.073,
|
| 315 |
+
"step": 2550
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 371.43,
|
| 319 |
+
"learning_rate": 6.913095238095239e-05,
|
| 320 |
+
"loss": 0.0627,
|
| 321 |
+
"step": 2600
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 378.57,
|
| 325 |
+
"learning_rate": 6.85357142857143e-05,
|
| 326 |
+
"loss": 0.0748,
|
| 327 |
+
"step": 2650
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 385.71,
|
| 331 |
+
"learning_rate": 6.79404761904762e-05,
|
| 332 |
+
"loss": 0.0668,
|
| 333 |
+
"step": 2700
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 392.86,
|
| 337 |
+
"learning_rate": 6.734523809523809e-05,
|
| 338 |
+
"loss": 0.0668,
|
| 339 |
+
"step": 2750
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 400.0,
|
| 343 |
+
"learning_rate": 6.675e-05,
|
| 344 |
+
"loss": 0.0813,
|
| 345 |
+
"step": 2800
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 407.14,
|
| 349 |
+
"learning_rate": 6.615476190476191e-05,
|
| 350 |
+
"loss": 0.0764,
|
| 351 |
+
"step": 2850
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 414.29,
|
| 355 |
+
"learning_rate": 6.555952380952382e-05,
|
| 356 |
+
"loss": 0.0671,
|
| 357 |
+
"step": 2900
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 421.43,
|
| 361 |
+
"learning_rate": 6.496428571428572e-05,
|
| 362 |
+
"loss": 0.0523,
|
| 363 |
+
"step": 2950
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 428.57,
|
| 367 |
+
"learning_rate": 6.436904761904761e-05,
|
| 368 |
+
"loss": 0.051,
|
| 369 |
+
"step": 3000
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 435.71,
|
| 373 |
+
"learning_rate": 6.377380952380952e-05,
|
| 374 |
+
"loss": 0.0612,
|
| 375 |
+
"step": 3050
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 442.86,
|
| 379 |
+
"learning_rate": 6.317857142857144e-05,
|
| 380 |
+
"loss": 0.0767,
|
| 381 |
+
"step": 3100
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 450.0,
|
| 385 |
+
"learning_rate": 6.258333333333334e-05,
|
| 386 |
+
"loss": 0.062,
|
| 387 |
+
"step": 3150
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 457.14,
|
| 391 |
+
"learning_rate": 6.198809523809525e-05,
|
| 392 |
+
"loss": 0.0704,
|
| 393 |
+
"step": 3200
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 464.29,
|
| 397 |
+
"learning_rate": 6.139285714285714e-05,
|
| 398 |
+
"loss": 0.0691,
|
| 399 |
+
"step": 3250
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 471.43,
|
| 403 |
+
"learning_rate": 6.079761904761905e-05,
|
| 404 |
+
"loss": 0.0654,
|
| 405 |
+
"step": 3300
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 478.57,
|
| 409 |
+
"learning_rate": 6.020238095238095e-05,
|
| 410 |
+
"loss": 0.055,
|
| 411 |
+
"step": 3350
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 485.71,
|
| 415 |
+
"learning_rate": 5.9607142857142864e-05,
|
| 416 |
+
"loss": 0.0572,
|
| 417 |
+
"step": 3400
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 492.86,
|
| 421 |
+
"learning_rate": 5.901190476190477e-05,
|
| 422 |
+
"loss": 0.0514,
|
| 423 |
+
"step": 3450
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 500.0,
|
| 427 |
+
"learning_rate": 5.8416666666666666e-05,
|
| 428 |
+
"loss": 0.0588,
|
| 429 |
+
"step": 3500
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 507.14,
|
| 433 |
+
"learning_rate": 5.782142857142857e-05,
|
| 434 |
+
"loss": 0.0612,
|
| 435 |
+
"step": 3550
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 514.29,
|
| 439 |
+
"learning_rate": 5.7226190476190475e-05,
|
| 440 |
+
"loss": 0.0629,
|
| 441 |
+
"step": 3600
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 521.43,
|
| 445 |
+
"learning_rate": 5.6630952380952386e-05,
|
| 446 |
+
"loss": 0.0558,
|
| 447 |
+
"step": 3650
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 528.57,
|
| 451 |
+
"learning_rate": 5.603571428571429e-05,
|
| 452 |
+
"loss": 0.0515,
|
| 453 |
+
"step": 3700
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 535.71,
|
| 457 |
+
"learning_rate": 5.5440476190476195e-05,
|
| 458 |
+
"loss": 0.0532,
|
| 459 |
+
"step": 3750
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 542.86,
|
| 463 |
+
"learning_rate": 5.484523809523809e-05,
|
| 464 |
+
"loss": 0.051,
|
| 465 |
+
"step": 3800
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 550.0,
|
| 469 |
+
"learning_rate": 5.4250000000000004e-05,
|
| 470 |
+
"loss": 0.0525,
|
| 471 |
+
"step": 3850
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 557.14,
|
| 475 |
+
"learning_rate": 5.365476190476191e-05,
|
| 476 |
+
"loss": 0.0465,
|
| 477 |
+
"step": 3900
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 564.29,
|
| 481 |
+
"learning_rate": 5.305952380952381e-05,
|
| 482 |
+
"loss": 0.0554,
|
| 483 |
+
"step": 3950
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 571.43,
|
| 487 |
+
"learning_rate": 5.2464285714285724e-05,
|
| 488 |
+
"loss": 0.0546,
|
| 489 |
+
"step": 4000
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 578.57,
|
| 493 |
+
"learning_rate": 5.1880952380952386e-05,
|
| 494 |
+
"loss": 0.0527,
|
| 495 |
+
"step": 4050
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 585.71,
|
| 499 |
+
"learning_rate": 5.128571428571429e-05,
|
| 500 |
+
"loss": 0.0565,
|
| 501 |
+
"step": 4100
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 592.86,
|
| 505 |
+
"learning_rate": 5.069047619047619e-05,
|
| 506 |
+
"loss": 0.051,
|
| 507 |
+
"step": 4150
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 600.0,
|
| 511 |
+
"learning_rate": 5.009523809523809e-05,
|
| 512 |
+
"loss": 0.0517,
|
| 513 |
+
"step": 4200
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 607.14,
|
| 517 |
+
"learning_rate": 4.9500000000000004e-05,
|
| 518 |
+
"loss": 0.0466,
|
| 519 |
+
"step": 4250
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 614.29,
|
| 523 |
+
"learning_rate": 4.890476190476191e-05,
|
| 524 |
+
"loss": 0.0558,
|
| 525 |
+
"step": 4300
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 621.43,
|
| 529 |
+
"learning_rate": 4.830952380952381e-05,
|
| 530 |
+
"loss": 0.055,
|
| 531 |
+
"step": 4350
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 628.57,
|
| 535 |
+
"learning_rate": 4.771428571428572e-05,
|
| 536 |
+
"loss": 0.0587,
|
| 537 |
+
"step": 4400
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 635.71,
|
| 541 |
+
"learning_rate": 4.711904761904762e-05,
|
| 542 |
+
"loss": 0.0545,
|
| 543 |
+
"step": 4450
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 642.86,
|
| 547 |
+
"learning_rate": 4.6523809523809526e-05,
|
| 548 |
+
"loss": 0.0454,
|
| 549 |
+
"step": 4500
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 650.0,
|
| 553 |
+
"learning_rate": 4.592857142857143e-05,
|
| 554 |
+
"loss": 0.0465,
|
| 555 |
+
"step": 4550
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 657.14,
|
| 559 |
+
"learning_rate": 4.5333333333333335e-05,
|
| 560 |
+
"loss": 0.0528,
|
| 561 |
+
"step": 4600
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 664.29,
|
| 565 |
+
"learning_rate": 4.473809523809524e-05,
|
| 566 |
+
"loss": 0.0584,
|
| 567 |
+
"step": 4650
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 671.43,
|
| 571 |
+
"learning_rate": 4.4142857142857144e-05,
|
| 572 |
+
"loss": 0.0466,
|
| 573 |
+
"step": 4700
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 678.57,
|
| 577 |
+
"learning_rate": 4.354761904761905e-05,
|
| 578 |
+
"loss": 0.0576,
|
| 579 |
+
"step": 4750
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 685.71,
|
| 583 |
+
"learning_rate": 4.295238095238095e-05,
|
| 584 |
+
"loss": 0.0496,
|
| 585 |
+
"step": 4800
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 692.86,
|
| 589 |
+
"learning_rate": 4.2357142857142864e-05,
|
| 590 |
+
"loss": 0.0401,
|
| 591 |
+
"step": 4850
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 700.0,
|
| 595 |
+
"learning_rate": 4.176190476190476e-05,
|
| 596 |
+
"loss": 0.0655,
|
| 597 |
+
"step": 4900
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 707.14,
|
| 601 |
+
"learning_rate": 4.116666666666667e-05,
|
| 602 |
+
"loss": 0.0533,
|
| 603 |
+
"step": 4950
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 714.29,
|
| 607 |
+
"learning_rate": 4.057142857142857e-05,
|
| 608 |
+
"loss": 0.0417,
|
| 609 |
+
"step": 5000
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 721.43,
|
| 613 |
+
"learning_rate": 3.9976190476190476e-05,
|
| 614 |
+
"loss": 0.0461,
|
| 615 |
+
"step": 5050
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 728.57,
|
| 619 |
+
"learning_rate": 3.938095238095239e-05,
|
| 620 |
+
"loss": 0.0434,
|
| 621 |
+
"step": 5100
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 735.71,
|
| 625 |
+
"learning_rate": 3.8785714285714285e-05,
|
| 626 |
+
"loss": 0.0409,
|
| 627 |
+
"step": 5150
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 742.86,
|
| 631 |
+
"learning_rate": 3.8190476190476196e-05,
|
| 632 |
+
"loss": 0.0404,
|
| 633 |
+
"step": 5200
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 750.0,
|
| 637 |
+
"learning_rate": 3.7595238095238094e-05,
|
| 638 |
+
"loss": 0.0494,
|
| 639 |
+
"step": 5250
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 757.14,
|
| 643 |
+
"learning_rate": 3.7e-05,
|
| 644 |
+
"loss": 0.0473,
|
| 645 |
+
"step": 5300
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 764.29,
|
| 649 |
+
"learning_rate": 3.640476190476191e-05,
|
| 650 |
+
"loss": 0.0449,
|
| 651 |
+
"step": 5350
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 771.43,
|
| 655 |
+
"learning_rate": 3.580952380952381e-05,
|
| 656 |
+
"loss": 0.044,
|
| 657 |
+
"step": 5400
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 778.57,
|
| 661 |
+
"learning_rate": 3.521428571428572e-05,
|
| 662 |
+
"loss": 0.0346,
|
| 663 |
+
"step": 5450
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 785.71,
|
| 667 |
+
"learning_rate": 3.461904761904762e-05,
|
| 668 |
+
"loss": 0.0436,
|
| 669 |
+
"step": 5500
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 792.86,
|
| 673 |
+
"learning_rate": 3.402380952380953e-05,
|
| 674 |
+
"loss": 0.0516,
|
| 675 |
+
"step": 5550
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 800.0,
|
| 679 |
+
"learning_rate": 3.342857142857143e-05,
|
| 680 |
+
"loss": 0.043,
|
| 681 |
+
"step": 5600
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 807.14,
|
| 685 |
+
"learning_rate": 3.283333333333333e-05,
|
| 686 |
+
"loss": 0.0494,
|
| 687 |
+
"step": 5650
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 814.29,
|
| 691 |
+
"learning_rate": 3.223809523809524e-05,
|
| 692 |
+
"loss": 0.0417,
|
| 693 |
+
"step": 5700
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 821.43,
|
| 697 |
+
"learning_rate": 3.1642857142857145e-05,
|
| 698 |
+
"loss": 0.0407,
|
| 699 |
+
"step": 5750
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 828.57,
|
| 703 |
+
"learning_rate": 3.104761904761905e-05,
|
| 704 |
+
"loss": 0.0387,
|
| 705 |
+
"step": 5800
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 835.71,
|
| 709 |
+
"learning_rate": 3.0452380952380954e-05,
|
| 710 |
+
"loss": 0.0418,
|
| 711 |
+
"step": 5850
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 842.86,
|
| 715 |
+
"learning_rate": 2.9857142857142862e-05,
|
| 716 |
+
"loss": 0.0407,
|
| 717 |
+
"step": 5900
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 850.0,
|
| 721 |
+
"learning_rate": 2.9261904761904763e-05,
|
| 722 |
+
"loss": 0.0348,
|
| 723 |
+
"step": 5950
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 857.14,
|
| 727 |
+
"learning_rate": 2.8666666666666668e-05,
|
| 728 |
+
"loss": 0.037,
|
| 729 |
+
"step": 6000
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 864.29,
|
| 733 |
+
"learning_rate": 2.8083333333333333e-05,
|
| 734 |
+
"loss": 0.041,
|
| 735 |
+
"step": 6050
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 871.43,
|
| 739 |
+
"learning_rate": 2.748809523809524e-05,
|
| 740 |
+
"loss": 0.0378,
|
| 741 |
+
"step": 6100
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 878.57,
|
| 745 |
+
"learning_rate": 2.689285714285714e-05,
|
| 746 |
+
"loss": 0.0364,
|
| 747 |
+
"step": 6150
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 885.71,
|
| 751 |
+
"learning_rate": 2.629761904761905e-05,
|
| 752 |
+
"loss": 0.0361,
|
| 753 |
+
"step": 6200
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 892.86,
|
| 757 |
+
"learning_rate": 2.5702380952380954e-05,
|
| 758 |
+
"loss": 0.0377,
|
| 759 |
+
"step": 6250
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 900.0,
|
| 763 |
+
"learning_rate": 2.510714285714286e-05,
|
| 764 |
+
"loss": 0.0353,
|
| 765 |
+
"step": 6300
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 907.14,
|
| 769 |
+
"learning_rate": 2.4511904761904763e-05,
|
| 770 |
+
"loss": 0.0323,
|
| 771 |
+
"step": 6350
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 914.29,
|
| 775 |
+
"learning_rate": 2.3916666666666668e-05,
|
| 776 |
+
"loss": 0.0413,
|
| 777 |
+
"step": 6400
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 921.43,
|
| 781 |
+
"learning_rate": 2.3321428571428572e-05,
|
| 782 |
+
"loss": 0.0358,
|
| 783 |
+
"step": 6450
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 928.57,
|
| 787 |
+
"learning_rate": 2.2726190476190476e-05,
|
| 788 |
+
"loss": 0.0367,
|
| 789 |
+
"step": 6500
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 935.71,
|
| 793 |
+
"learning_rate": 2.2130952380952384e-05,
|
| 794 |
+
"loss": 0.0401,
|
| 795 |
+
"step": 6550
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 942.86,
|
| 799 |
+
"learning_rate": 2.1535714285714285e-05,
|
| 800 |
+
"loss": 0.0344,
|
| 801 |
+
"step": 6600
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 950.0,
|
| 805 |
+
"learning_rate": 2.094047619047619e-05,
|
| 806 |
+
"loss": 0.0297,
|
| 807 |
+
"step": 6650
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 957.14,
|
| 811 |
+
"learning_rate": 2.0345238095238094e-05,
|
| 812 |
+
"loss": 0.0292,
|
| 813 |
+
"step": 6700
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 964.29,
|
| 817 |
+
"learning_rate": 1.9750000000000002e-05,
|
| 818 |
+
"loss": 0.0334,
|
| 819 |
+
"step": 6750
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 971.43,
|
| 823 |
+
"learning_rate": 1.9154761904761907e-05,
|
| 824 |
+
"loss": 0.0378,
|
| 825 |
+
"step": 6800
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 978.57,
|
| 829 |
+
"learning_rate": 1.855952380952381e-05,
|
| 830 |
+
"loss": 0.0386,
|
| 831 |
+
"step": 6850
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 985.71,
|
| 835 |
+
"learning_rate": 1.7964285714285712e-05,
|
| 836 |
+
"loss": 0.0284,
|
| 837 |
+
"step": 6900
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 992.86,
|
| 841 |
+
"learning_rate": 1.736904761904762e-05,
|
| 842 |
+
"loss": 0.0349,
|
| 843 |
+
"step": 6950
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 1000.0,
|
| 847 |
+
"learning_rate": 1.6773809523809525e-05,
|
| 848 |
+
"loss": 0.0296,
|
| 849 |
+
"step": 7000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 1007.14,
|
| 853 |
+
"learning_rate": 1.617857142857143e-05,
|
| 854 |
+
"loss": 0.0302,
|
| 855 |
+
"step": 7050
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 1014.29,
|
| 859 |
+
"learning_rate": 1.5583333333333334e-05,
|
| 860 |
+
"loss": 0.0265,
|
| 861 |
+
"step": 7100
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 1021.43,
|
| 865 |
+
"learning_rate": 1.498809523809524e-05,
|
| 866 |
+
"loss": 0.0288,
|
| 867 |
+
"step": 7150
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 1028.57,
|
| 871 |
+
"learning_rate": 1.4392857142857144e-05,
|
| 872 |
+
"loss": 0.0318,
|
| 873 |
+
"step": 7200
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 1035.71,
|
| 877 |
+
"learning_rate": 1.3797619047619049e-05,
|
| 878 |
+
"loss": 0.0286,
|
| 879 |
+
"step": 7250
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 1042.86,
|
| 883 |
+
"learning_rate": 1.3202380952380952e-05,
|
| 884 |
+
"loss": 0.0328,
|
| 885 |
+
"step": 7300
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 1050.0,
|
| 889 |
+
"learning_rate": 1.260714285714286e-05,
|
| 890 |
+
"loss": 0.027,
|
| 891 |
+
"step": 7350
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 1057.14,
|
| 895 |
+
"learning_rate": 1.2011904761904762e-05,
|
| 896 |
+
"loss": 0.0266,
|
| 897 |
+
"step": 7400
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1064.29,
|
| 901 |
+
"learning_rate": 1.1416666666666667e-05,
|
| 902 |
+
"loss": 0.0271,
|
| 903 |
+
"step": 7450
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 1071.43,
|
| 907 |
+
"learning_rate": 1.0821428571428573e-05,
|
| 908 |
+
"loss": 0.0252,
|
| 909 |
+
"step": 7500
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 1078.57,
|
| 913 |
+
"learning_rate": 1.0226190476190476e-05,
|
| 914 |
+
"loss": 0.0264,
|
| 915 |
+
"step": 7550
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 1085.71,
|
| 919 |
+
"learning_rate": 9.63095238095238e-06,
|
| 920 |
+
"loss": 0.0238,
|
| 921 |
+
"step": 7600
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 1092.86,
|
| 925 |
+
"learning_rate": 9.035714285714287e-06,
|
| 926 |
+
"loss": 0.0276,
|
| 927 |
+
"step": 7650
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 1100.0,
|
| 931 |
+
"learning_rate": 8.440476190476191e-06,
|
| 932 |
+
"loss": 0.026,
|
| 933 |
+
"step": 7700
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 1107.14,
|
| 937 |
+
"learning_rate": 7.845238095238096e-06,
|
| 938 |
+
"loss": 0.0253,
|
| 939 |
+
"step": 7750
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 1114.29,
|
| 943 |
+
"learning_rate": 7.25e-06,
|
| 944 |
+
"loss": 0.022,
|
| 945 |
+
"step": 7800
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 1121.43,
|
| 949 |
+
"learning_rate": 6.654761904761905e-06,
|
| 950 |
+
"loss": 0.0219,
|
| 951 |
+
"step": 7850
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 1128.57,
|
| 955 |
+
"learning_rate": 6.05952380952381e-06,
|
| 956 |
+
"loss": 0.0212,
|
| 957 |
+
"step": 7900
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 1135.71,
|
| 961 |
+
"learning_rate": 5.464285714285715e-06,
|
| 962 |
+
"loss": 0.0249,
|
| 963 |
+
"step": 7950
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 1142.86,
|
| 967 |
+
"learning_rate": 4.86904761904762e-06,
|
| 968 |
+
"loss": 0.0244,
|
| 969 |
+
"step": 8000
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 1150.0,
|
| 973 |
+
"learning_rate": 4.273809523809523e-06,
|
| 974 |
+
"loss": 0.023,
|
| 975 |
+
"step": 8050
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 1157.14,
|
| 979 |
+
"learning_rate": 3.690476190476191e-06,
|
| 980 |
+
"loss": 0.0227,
|
| 981 |
+
"step": 8100
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 1164.29,
|
| 985 |
+
"learning_rate": 3.0952380952380953e-06,
|
| 986 |
+
"loss": 0.0244,
|
| 987 |
+
"step": 8150
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 1171.43,
|
| 991 |
+
"learning_rate": 2.5e-06,
|
| 992 |
+
"loss": 0.0205,
|
| 993 |
+
"step": 8200
|
| 994 |
+
}
|
| 995 |
+
],
|
| 996 |
+
"logging_steps": 50,
|
| 997 |
+
"max_steps": 8400,
|
| 998 |
+
"num_train_epochs": 1200,
|
| 999 |
+
"save_steps": 200,
|
| 1000 |
+
"total_flos": 6.852747656247149e+18,
|
| 1001 |
+
"trial_name": null,
|
| 1002 |
+
"trial_params": null
|
| 1003 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4268084a8124ca35e5140aa8ae9bf43f20b3fd8968ea2eda95807cf778c56d61
|
| 3 |
+
size 4027
|