Commit
·
36de9bd
1
Parent(s):
414537c
Upload 8 files
Browse files- config.json +56 -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 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "microsoft/table-transformer-structure-recognition",
|
| 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": "header",
|
| 33 |
+
"1": "line",
|
| 34 |
+
"2": "column"
|
| 35 |
+
},
|
| 36 |
+
"init_std": 0.02,
|
| 37 |
+
"init_xavier_std": 1.0,
|
| 38 |
+
"is_encoder_decoder": true,
|
| 39 |
+
"label2id": {
|
| 40 |
+
"column": "2",
|
| 41 |
+
"header": "0",
|
| 42 |
+
"line": "1"
|
| 43 |
+
},
|
| 44 |
+
"mask_loss_coefficient": 1,
|
| 45 |
+
"max_position_embeddings": 1024,
|
| 46 |
+
"model_type": "table-transformer",
|
| 47 |
+
"num_channels": 3,
|
| 48 |
+
"num_hidden_layers": 6,
|
| 49 |
+
"num_queries": 125,
|
| 50 |
+
"position_embedding_type": "sine",
|
| 51 |
+
"scale_embedding": false,
|
| 52 |
+
"torch_dtype": "float32",
|
| 53 |
+
"transformers_version": "4.33.1",
|
| 54 |
+
"use_pretrained_backbone": true,
|
| 55 |
+
"use_timm_backbone": true
|
| 56 |
+
}
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8093f849a1f44f636f51ea7eb018488f211d1b7ea335db3098421681be564d9f
|
| 3 |
+
size 229603589
|
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 sha256:5990464b6be1bd55bbde3b0de81dc23c52e9b576972b8b78f259c9be419ecc6a
|
| 3 |
+
size 115513801
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7988245f19c1a0083ef5451b1b552ac5e490cf767e0e7b85b639fad0a0668442
|
| 3 |
+
size 14575
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:60a508a69f71066c3f15ed1c4d473afee88d83aad500533d07f8d5c9aada541e
|
| 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": 9.945238095238095e-05,
|
| 14 |
+
"loss": 2.1185,
|
| 15 |
+
"step": 50
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 14.29,
|
| 19 |
+
"learning_rate": 9.885714285714286e-05,
|
| 20 |
+
"loss": 1.3056,
|
| 21 |
+
"step": 100
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 21.43,
|
| 25 |
+
"learning_rate": 9.826190476190476e-05,
|
| 26 |
+
"loss": 0.9754,
|
| 27 |
+
"step": 150
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 28.57,
|
| 31 |
+
"learning_rate": 9.767857142857144e-05,
|
| 32 |
+
"loss": 0.7447,
|
| 33 |
+
"step": 200
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 35.71,
|
| 37 |
+
"learning_rate": 9.708333333333334e-05,
|
| 38 |
+
"loss": 0.6164,
|
| 39 |
+
"step": 250
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 42.86,
|
| 43 |
+
"learning_rate": 9.648809523809524e-05,
|
| 44 |
+
"loss": 0.5674,
|
| 45 |
+
"step": 300
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 50.0,
|
| 49 |
+
"learning_rate": 9.589285714285715e-05,
|
| 50 |
+
"loss": 0.5373,
|
| 51 |
+
"step": 350
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 57.14,
|
| 55 |
+
"learning_rate": 9.529761904761905e-05,
|
| 56 |
+
"loss": 0.5459,
|
| 57 |
+
"step": 400
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 64.29,
|
| 61 |
+
"learning_rate": 9.470238095238096e-05,
|
| 62 |
+
"loss": 0.4995,
|
| 63 |
+
"step": 450
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 71.43,
|
| 67 |
+
"learning_rate": 9.410714285714286e-05,
|
| 68 |
+
"loss": 0.4877,
|
| 69 |
+
"step": 500
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 78.57,
|
| 73 |
+
"learning_rate": 9.351190476190477e-05,
|
| 74 |
+
"loss": 0.5287,
|
| 75 |
+
"step": 550
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 85.71,
|
| 79 |
+
"learning_rate": 9.291666666666667e-05,
|
| 80 |
+
"loss": 0.4737,
|
| 81 |
+
"step": 600
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 92.86,
|
| 85 |
+
"learning_rate": 9.232142857142858e-05,
|
| 86 |
+
"loss": 0.4594,
|
| 87 |
+
"step": 650
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 100.0,
|
| 91 |
+
"learning_rate": 9.172619047619048e-05,
|
| 92 |
+
"loss": 0.4416,
|
| 93 |
+
"step": 700
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 107.14,
|
| 97 |
+
"learning_rate": 9.113095238095238e-05,
|
| 98 |
+
"loss": 0.446,
|
| 99 |
+
"step": 750
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 114.29,
|
| 103 |
+
"learning_rate": 9.053571428571429e-05,
|
| 104 |
+
"loss": 0.4313,
|
| 105 |
+
"step": 800
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 121.43,
|
| 109 |
+
"learning_rate": 8.99404761904762e-05,
|
| 110 |
+
"loss": 0.421,
|
| 111 |
+
"step": 850
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 128.57,
|
| 115 |
+
"learning_rate": 8.93452380952381e-05,
|
| 116 |
+
"loss": 0.4148,
|
| 117 |
+
"step": 900
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 135.71,
|
| 121 |
+
"learning_rate": 8.875e-05,
|
| 122 |
+
"loss": 0.4191,
|
| 123 |
+
"step": 950
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 142.86,
|
| 127 |
+
"learning_rate": 8.815476190476191e-05,
|
| 128 |
+
"loss": 0.4013,
|
| 129 |
+
"step": 1000
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 150.0,
|
| 133 |
+
"learning_rate": 8.755952380952381e-05,
|
| 134 |
+
"loss": 0.4001,
|
| 135 |
+
"step": 1050
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 157.14,
|
| 139 |
+
"learning_rate": 8.696428571428572e-05,
|
| 140 |
+
"loss": 0.3944,
|
| 141 |
+
"step": 1100
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 164.29,
|
| 145 |
+
"learning_rate": 8.636904761904763e-05,
|
| 146 |
+
"loss": 0.3911,
|
| 147 |
+
"step": 1150
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 171.43,
|
| 151 |
+
"learning_rate": 8.577380952380952e-05,
|
| 152 |
+
"loss": 0.3679,
|
| 153 |
+
"step": 1200
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 178.57,
|
| 157 |
+
"learning_rate": 8.517857142857143e-05,
|
| 158 |
+
"loss": 0.3749,
|
| 159 |
+
"step": 1250
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 185.71,
|
| 163 |
+
"learning_rate": 8.458333333333333e-05,
|
| 164 |
+
"loss": 0.3781,
|
| 165 |
+
"step": 1300
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 192.86,
|
| 169 |
+
"learning_rate": 8.398809523809524e-05,
|
| 170 |
+
"loss": 0.3628,
|
| 171 |
+
"step": 1350
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 200.0,
|
| 175 |
+
"learning_rate": 8.339285714285716e-05,
|
| 176 |
+
"loss": 0.3614,
|
| 177 |
+
"step": 1400
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 207.14,
|
| 181 |
+
"learning_rate": 8.279761904761905e-05,
|
| 182 |
+
"loss": 0.3433,
|
| 183 |
+
"step": 1450
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 214.29,
|
| 187 |
+
"learning_rate": 8.220238095238095e-05,
|
| 188 |
+
"loss": 0.331,
|
| 189 |
+
"step": 1500
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 221.43,
|
| 193 |
+
"learning_rate": 8.160714285714286e-05,
|
| 194 |
+
"loss": 0.3128,
|
| 195 |
+
"step": 1550
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 228.57,
|
| 199 |
+
"learning_rate": 8.101190476190476e-05,
|
| 200 |
+
"loss": 0.3186,
|
| 201 |
+
"step": 1600
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 235.71,
|
| 205 |
+
"learning_rate": 8.041666666666668e-05,
|
| 206 |
+
"loss": 0.2911,
|
| 207 |
+
"step": 1650
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 242.86,
|
| 211 |
+
"learning_rate": 7.982142857142858e-05,
|
| 212 |
+
"loss": 0.2851,
|
| 213 |
+
"step": 1700
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 250.0,
|
| 217 |
+
"learning_rate": 7.922619047619047e-05,
|
| 218 |
+
"loss": 0.2748,
|
| 219 |
+
"step": 1750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 257.14,
|
| 223 |
+
"learning_rate": 7.863095238095238e-05,
|
| 224 |
+
"loss": 0.2614,
|
| 225 |
+
"step": 1800
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 264.29,
|
| 229 |
+
"learning_rate": 7.803571428571428e-05,
|
| 230 |
+
"loss": 0.2631,
|
| 231 |
+
"step": 1850
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 271.43,
|
| 235 |
+
"learning_rate": 7.74404761904762e-05,
|
| 236 |
+
"loss": 0.2467,
|
| 237 |
+
"step": 1900
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 278.57,
|
| 241 |
+
"learning_rate": 7.68452380952381e-05,
|
| 242 |
+
"loss": 0.2372,
|
| 243 |
+
"step": 1950
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 285.71,
|
| 247 |
+
"learning_rate": 7.625e-05,
|
| 248 |
+
"loss": 0.23,
|
| 249 |
+
"step": 2000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 292.86,
|
| 253 |
+
"learning_rate": 7.56547619047619e-05,
|
| 254 |
+
"loss": 0.2314,
|
| 255 |
+
"step": 2050
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 300.0,
|
| 259 |
+
"learning_rate": 7.505952380952382e-05,
|
| 260 |
+
"loss": 0.2218,
|
| 261 |
+
"step": 2100
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 307.14,
|
| 265 |
+
"learning_rate": 7.446428571428572e-05,
|
| 266 |
+
"loss": 0.2165,
|
| 267 |
+
"step": 2150
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 314.29,
|
| 271 |
+
"learning_rate": 7.386904761904763e-05,
|
| 272 |
+
"loss": 0.2262,
|
| 273 |
+
"step": 2200
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 321.43,
|
| 277 |
+
"learning_rate": 7.327380952380952e-05,
|
| 278 |
+
"loss": 0.22,
|
| 279 |
+
"step": 2250
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 328.57,
|
| 283 |
+
"learning_rate": 7.267857142857142e-05,
|
| 284 |
+
"loss": 0.2099,
|
| 285 |
+
"step": 2300
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 335.71,
|
| 289 |
+
"learning_rate": 7.208333333333334e-05,
|
| 290 |
+
"loss": 0.2138,
|
| 291 |
+
"step": 2350
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 342.86,
|
| 295 |
+
"learning_rate": 7.148809523809525e-05,
|
| 296 |
+
"loss": 0.2064,
|
| 297 |
+
"step": 2400
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 350.0,
|
| 301 |
+
"learning_rate": 7.089285714285715e-05,
|
| 302 |
+
"loss": 0.197,
|
| 303 |
+
"step": 2450
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 357.14,
|
| 307 |
+
"learning_rate": 7.030952380952381e-05,
|
| 308 |
+
"loss": 0.191,
|
| 309 |
+
"step": 2500
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 364.29,
|
| 313 |
+
"learning_rate": 6.971428571428572e-05,
|
| 314 |
+
"loss": 0.1771,
|
| 315 |
+
"step": 2550
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 371.43,
|
| 319 |
+
"learning_rate": 6.911904761904762e-05,
|
| 320 |
+
"loss": 0.1912,
|
| 321 |
+
"step": 2600
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 378.57,
|
| 325 |
+
"learning_rate": 6.852380952380953e-05,
|
| 326 |
+
"loss": 0.1597,
|
| 327 |
+
"step": 2650
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 385.71,
|
| 331 |
+
"learning_rate": 6.792857142857143e-05,
|
| 332 |
+
"loss": 0.1584,
|
| 333 |
+
"step": 2700
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 392.86,
|
| 337 |
+
"learning_rate": 6.733333333333333e-05,
|
| 338 |
+
"loss": 0.1518,
|
| 339 |
+
"step": 2750
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 400.0,
|
| 343 |
+
"learning_rate": 6.673809523809524e-05,
|
| 344 |
+
"loss": 0.1492,
|
| 345 |
+
"step": 2800
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 407.14,
|
| 349 |
+
"learning_rate": 6.614285714285716e-05,
|
| 350 |
+
"loss": 0.1609,
|
| 351 |
+
"step": 2850
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 414.29,
|
| 355 |
+
"learning_rate": 6.554761904761905e-05,
|
| 356 |
+
"loss": 0.1424,
|
| 357 |
+
"step": 2900
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 421.43,
|
| 361 |
+
"learning_rate": 6.495238095238095e-05,
|
| 362 |
+
"loss": 0.1405,
|
| 363 |
+
"step": 2950
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 428.57,
|
| 367 |
+
"learning_rate": 6.435714285714286e-05,
|
| 368 |
+
"loss": 0.1352,
|
| 369 |
+
"step": 3000
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 435.71,
|
| 373 |
+
"learning_rate": 6.376190476190476e-05,
|
| 374 |
+
"loss": 0.1286,
|
| 375 |
+
"step": 3050
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 442.86,
|
| 379 |
+
"learning_rate": 6.316666666666668e-05,
|
| 380 |
+
"loss": 0.1243,
|
| 381 |
+
"step": 3100
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 450.0,
|
| 385 |
+
"learning_rate": 6.257142857142857e-05,
|
| 386 |
+
"loss": 0.1237,
|
| 387 |
+
"step": 3150
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 457.14,
|
| 391 |
+
"learning_rate": 6.197619047619048e-05,
|
| 392 |
+
"loss": 0.1232,
|
| 393 |
+
"step": 3200
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 464.29,
|
| 397 |
+
"learning_rate": 6.138095238095238e-05,
|
| 398 |
+
"loss": 0.1226,
|
| 399 |
+
"step": 3250
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 471.43,
|
| 403 |
+
"learning_rate": 6.078571428571429e-05,
|
| 404 |
+
"loss": 0.1147,
|
| 405 |
+
"step": 3300
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 478.57,
|
| 409 |
+
"learning_rate": 6.0190476190476195e-05,
|
| 410 |
+
"loss": 0.1161,
|
| 411 |
+
"step": 3350
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 485.71,
|
| 415 |
+
"learning_rate": 5.959523809523809e-05,
|
| 416 |
+
"loss": 0.1208,
|
| 417 |
+
"step": 3400
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 492.86,
|
| 421 |
+
"learning_rate": 5.9e-05,
|
| 422 |
+
"loss": 0.1224,
|
| 423 |
+
"step": 3450
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 500.0,
|
| 427 |
+
"learning_rate": 5.840476190476191e-05,
|
| 428 |
+
"loss": 0.1116,
|
| 429 |
+
"step": 3500
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 507.14,
|
| 433 |
+
"learning_rate": 5.780952380952381e-05,
|
| 434 |
+
"loss": 0.1074,
|
| 435 |
+
"step": 3550
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 514.29,
|
| 439 |
+
"learning_rate": 5.721428571428572e-05,
|
| 440 |
+
"loss": 0.1036,
|
| 441 |
+
"step": 3600
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 521.43,
|
| 445 |
+
"learning_rate": 5.661904761904763e-05,
|
| 446 |
+
"loss": 0.1074,
|
| 447 |
+
"step": 3650
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 528.57,
|
| 451 |
+
"learning_rate": 5.602380952380952e-05,
|
| 452 |
+
"loss": 0.1042,
|
| 453 |
+
"step": 3700
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 535.71,
|
| 457 |
+
"learning_rate": 5.542857142857143e-05,
|
| 458 |
+
"loss": 0.1074,
|
| 459 |
+
"step": 3750
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 542.86,
|
| 463 |
+
"learning_rate": 5.4833333333333336e-05,
|
| 464 |
+
"loss": 0.1033,
|
| 465 |
+
"step": 3800
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 550.0,
|
| 469 |
+
"learning_rate": 5.423809523809524e-05,
|
| 470 |
+
"loss": 0.0996,
|
| 471 |
+
"step": 3850
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 557.14,
|
| 475 |
+
"learning_rate": 5.364285714285715e-05,
|
| 476 |
+
"loss": 0.0969,
|
| 477 |
+
"step": 3900
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 564.29,
|
| 481 |
+
"learning_rate": 5.304761904761905e-05,
|
| 482 |
+
"loss": 0.0943,
|
| 483 |
+
"step": 3950
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 571.43,
|
| 487 |
+
"learning_rate": 5.2452380952380954e-05,
|
| 488 |
+
"loss": 0.1012,
|
| 489 |
+
"step": 4000
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 578.57,
|
| 493 |
+
"learning_rate": 5.185714285714286e-05,
|
| 494 |
+
"loss": 0.1011,
|
| 495 |
+
"step": 4050
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 585.71,
|
| 499 |
+
"learning_rate": 5.126190476190477e-05,
|
| 500 |
+
"loss": 0.0981,
|
| 501 |
+
"step": 4100
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 592.86,
|
| 505 |
+
"learning_rate": 5.0666666666666674e-05,
|
| 506 |
+
"loss": 0.0935,
|
| 507 |
+
"step": 4150
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 600.0,
|
| 511 |
+
"learning_rate": 5.007142857142857e-05,
|
| 512 |
+
"loss": 0.0922,
|
| 513 |
+
"step": 4200
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 607.14,
|
| 517 |
+
"learning_rate": 4.9476190476190476e-05,
|
| 518 |
+
"loss": 0.0935,
|
| 519 |
+
"step": 4250
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 614.29,
|
| 523 |
+
"learning_rate": 4.888095238095238e-05,
|
| 524 |
+
"loss": 0.0955,
|
| 525 |
+
"step": 4300
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 621.43,
|
| 529 |
+
"learning_rate": 4.828571428571429e-05,
|
| 530 |
+
"loss": 0.0909,
|
| 531 |
+
"step": 4350
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 628.57,
|
| 535 |
+
"learning_rate": 4.769047619047619e-05,
|
| 536 |
+
"loss": 0.0886,
|
| 537 |
+
"step": 4400
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 635.71,
|
| 541 |
+
"learning_rate": 4.70952380952381e-05,
|
| 542 |
+
"loss": 0.0886,
|
| 543 |
+
"step": 4450
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 642.86,
|
| 547 |
+
"learning_rate": 4.6500000000000005e-05,
|
| 548 |
+
"loss": 0.0884,
|
| 549 |
+
"step": 4500
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 650.0,
|
| 553 |
+
"learning_rate": 4.591666666666667e-05,
|
| 554 |
+
"loss": 0.0939,
|
| 555 |
+
"step": 4550
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 657.14,
|
| 559 |
+
"learning_rate": 4.532142857142858e-05,
|
| 560 |
+
"loss": 0.0879,
|
| 561 |
+
"step": 4600
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 664.29,
|
| 565 |
+
"learning_rate": 4.4726190476190476e-05,
|
| 566 |
+
"loss": 0.0867,
|
| 567 |
+
"step": 4650
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 671.43,
|
| 571 |
+
"learning_rate": 4.413095238095239e-05,
|
| 572 |
+
"loss": 0.0858,
|
| 573 |
+
"step": 4700
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 678.57,
|
| 577 |
+
"learning_rate": 4.3535714285714285e-05,
|
| 578 |
+
"loss": 0.0882,
|
| 579 |
+
"step": 4750
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 685.71,
|
| 583 |
+
"learning_rate": 4.294047619047619e-05,
|
| 584 |
+
"loss": 0.085,
|
| 585 |
+
"step": 4800
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 692.86,
|
| 589 |
+
"learning_rate": 4.23452380952381e-05,
|
| 590 |
+
"loss": 0.0787,
|
| 591 |
+
"step": 4850
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 700.0,
|
| 595 |
+
"learning_rate": 4.175e-05,
|
| 596 |
+
"loss": 0.0798,
|
| 597 |
+
"step": 4900
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 707.14,
|
| 601 |
+
"learning_rate": 4.115476190476191e-05,
|
| 602 |
+
"loss": 0.0805,
|
| 603 |
+
"step": 4950
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 714.29,
|
| 607 |
+
"learning_rate": 4.0559523809523814e-05,
|
| 608 |
+
"loss": 0.0796,
|
| 609 |
+
"step": 5000
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 721.43,
|
| 613 |
+
"learning_rate": 3.996428571428571e-05,
|
| 614 |
+
"loss": 0.084,
|
| 615 |
+
"step": 5050
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 728.57,
|
| 619 |
+
"learning_rate": 3.936904761904762e-05,
|
| 620 |
+
"loss": 0.0826,
|
| 621 |
+
"step": 5100
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 735.71,
|
| 625 |
+
"learning_rate": 3.877380952380952e-05,
|
| 626 |
+
"loss": 0.0807,
|
| 627 |
+
"step": 5150
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 742.86,
|
| 631 |
+
"learning_rate": 3.817857142857143e-05,
|
| 632 |
+
"loss": 0.0744,
|
| 633 |
+
"step": 5200
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 750.0,
|
| 637 |
+
"learning_rate": 3.7583333333333337e-05,
|
| 638 |
+
"loss": 0.0845,
|
| 639 |
+
"step": 5250
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 757.14,
|
| 643 |
+
"learning_rate": 3.698809523809524e-05,
|
| 644 |
+
"loss": 0.0785,
|
| 645 |
+
"step": 5300
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 764.29,
|
| 649 |
+
"learning_rate": 3.6392857142857146e-05,
|
| 650 |
+
"loss": 0.0786,
|
| 651 |
+
"step": 5350
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 771.43,
|
| 655 |
+
"learning_rate": 3.579761904761905e-05,
|
| 656 |
+
"loss": 0.0743,
|
| 657 |
+
"step": 5400
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 778.57,
|
| 661 |
+
"learning_rate": 3.5202380952380955e-05,
|
| 662 |
+
"loss": 0.0778,
|
| 663 |
+
"step": 5450
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 785.71,
|
| 667 |
+
"learning_rate": 3.460714285714286e-05,
|
| 668 |
+
"loss": 0.0706,
|
| 669 |
+
"step": 5500
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 792.86,
|
| 673 |
+
"learning_rate": 3.4011904761904764e-05,
|
| 674 |
+
"loss": 0.0735,
|
| 675 |
+
"step": 5550
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 800.0,
|
| 679 |
+
"learning_rate": 3.341666666666667e-05,
|
| 680 |
+
"loss": 0.0727,
|
| 681 |
+
"step": 5600
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 807.14,
|
| 685 |
+
"learning_rate": 3.282142857142857e-05,
|
| 686 |
+
"loss": 0.0779,
|
| 687 |
+
"step": 5650
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 814.29,
|
| 691 |
+
"learning_rate": 3.222619047619048e-05,
|
| 692 |
+
"loss": 0.0708,
|
| 693 |
+
"step": 5700
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 821.43,
|
| 697 |
+
"learning_rate": 3.163095238095238e-05,
|
| 698 |
+
"loss": 0.0708,
|
| 699 |
+
"step": 5750
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 828.57,
|
| 703 |
+
"learning_rate": 3.1035714285714286e-05,
|
| 704 |
+
"loss": 0.0709,
|
| 705 |
+
"step": 5800
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 835.71,
|
| 709 |
+
"learning_rate": 3.044047619047619e-05,
|
| 710 |
+
"loss": 0.0719,
|
| 711 |
+
"step": 5850
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 842.86,
|
| 715 |
+
"learning_rate": 2.98452380952381e-05,
|
| 716 |
+
"loss": 0.0721,
|
| 717 |
+
"step": 5900
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 850.0,
|
| 721 |
+
"learning_rate": 2.925e-05,
|
| 722 |
+
"loss": 0.0685,
|
| 723 |
+
"step": 5950
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 857.14,
|
| 727 |
+
"learning_rate": 2.8654761904761907e-05,
|
| 728 |
+
"loss": 0.0711,
|
| 729 |
+
"step": 6000
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 864.29,
|
| 733 |
+
"learning_rate": 2.8059523809523812e-05,
|
| 734 |
+
"loss": 0.0645,
|
| 735 |
+
"step": 6050
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 871.43,
|
| 739 |
+
"learning_rate": 2.7464285714285713e-05,
|
| 740 |
+
"loss": 0.0694,
|
| 741 |
+
"step": 6100
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 878.57,
|
| 745 |
+
"learning_rate": 2.686904761904762e-05,
|
| 746 |
+
"loss": 0.0636,
|
| 747 |
+
"step": 6150
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 885.71,
|
| 751 |
+
"learning_rate": 2.6273809523809522e-05,
|
| 752 |
+
"loss": 0.0635,
|
| 753 |
+
"step": 6200
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 892.86,
|
| 757 |
+
"learning_rate": 2.567857142857143e-05,
|
| 758 |
+
"loss": 0.0636,
|
| 759 |
+
"step": 6250
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 900.0,
|
| 763 |
+
"learning_rate": 2.5083333333333338e-05,
|
| 764 |
+
"loss": 0.0676,
|
| 765 |
+
"step": 6300
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 907.14,
|
| 769 |
+
"learning_rate": 2.4488095238095242e-05,
|
| 770 |
+
"loss": 0.0614,
|
| 771 |
+
"step": 6350
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 914.29,
|
| 775 |
+
"learning_rate": 2.3892857142857143e-05,
|
| 776 |
+
"loss": 0.0641,
|
| 777 |
+
"step": 6400
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 921.43,
|
| 781 |
+
"learning_rate": 2.3297619047619048e-05,
|
| 782 |
+
"loss": 0.0649,
|
| 783 |
+
"step": 6450
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 928.57,
|
| 787 |
+
"learning_rate": 2.2702380952380952e-05,
|
| 788 |
+
"loss": 0.0627,
|
| 789 |
+
"step": 6500
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 935.71,
|
| 793 |
+
"learning_rate": 2.2107142857142857e-05,
|
| 794 |
+
"loss": 0.062,
|
| 795 |
+
"step": 6550
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 942.86,
|
| 799 |
+
"learning_rate": 2.1511904761904765e-05,
|
| 800 |
+
"loss": 0.0621,
|
| 801 |
+
"step": 6600
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 950.0,
|
| 805 |
+
"learning_rate": 2.091666666666667e-05,
|
| 806 |
+
"loss": 0.061,
|
| 807 |
+
"step": 6650
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 957.14,
|
| 811 |
+
"learning_rate": 2.032142857142857e-05,
|
| 812 |
+
"loss": 0.0603,
|
| 813 |
+
"step": 6700
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 964.29,
|
| 817 |
+
"learning_rate": 1.9726190476190475e-05,
|
| 818 |
+
"loss": 0.0586,
|
| 819 |
+
"step": 6750
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 971.43,
|
| 823 |
+
"learning_rate": 1.9130952380952383e-05,
|
| 824 |
+
"loss": 0.0586,
|
| 825 |
+
"step": 6800
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 978.57,
|
| 829 |
+
"learning_rate": 1.8535714285714287e-05,
|
| 830 |
+
"loss": 0.0582,
|
| 831 |
+
"step": 6850
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 985.71,
|
| 835 |
+
"learning_rate": 1.794047619047619e-05,
|
| 836 |
+
"loss": 0.0586,
|
| 837 |
+
"step": 6900
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 992.86,
|
| 841 |
+
"learning_rate": 1.7345238095238096e-05,
|
| 842 |
+
"loss": 0.0553,
|
| 843 |
+
"step": 6950
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 1000.0,
|
| 847 |
+
"learning_rate": 1.675e-05,
|
| 848 |
+
"loss": 0.056,
|
| 849 |
+
"step": 7000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 1007.14,
|
| 853 |
+
"learning_rate": 1.6166666666666665e-05,
|
| 854 |
+
"loss": 0.0596,
|
| 855 |
+
"step": 7050
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 1014.29,
|
| 859 |
+
"learning_rate": 1.5571428571428573e-05,
|
| 860 |
+
"loss": 0.055,
|
| 861 |
+
"step": 7100
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 1021.43,
|
| 865 |
+
"learning_rate": 1.4976190476190476e-05,
|
| 866 |
+
"loss": 0.0545,
|
| 867 |
+
"step": 7150
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 1028.57,
|
| 871 |
+
"learning_rate": 1.438095238095238e-05,
|
| 872 |
+
"loss": 0.054,
|
| 873 |
+
"step": 7200
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 1035.71,
|
| 877 |
+
"learning_rate": 1.3785714285714285e-05,
|
| 878 |
+
"loss": 0.0543,
|
| 879 |
+
"step": 7250
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 1042.86,
|
| 883 |
+
"learning_rate": 1.3190476190476191e-05,
|
| 884 |
+
"loss": 0.0537,
|
| 885 |
+
"step": 7300
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 1050.0,
|
| 889 |
+
"learning_rate": 1.2595238095238096e-05,
|
| 890 |
+
"loss": 0.0553,
|
| 891 |
+
"step": 7350
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 1057.14,
|
| 895 |
+
"learning_rate": 1.2e-05,
|
| 896 |
+
"loss": 0.0532,
|
| 897 |
+
"step": 7400
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1064.29,
|
| 901 |
+
"learning_rate": 1.1404761904761905e-05,
|
| 902 |
+
"loss": 0.053,
|
| 903 |
+
"step": 7450
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 1071.43,
|
| 907 |
+
"learning_rate": 1.080952380952381e-05,
|
| 908 |
+
"loss": 0.0531,
|
| 909 |
+
"step": 7500
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 1078.57,
|
| 913 |
+
"learning_rate": 1.0214285714285715e-05,
|
| 914 |
+
"loss": 0.0546,
|
| 915 |
+
"step": 7550
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 1085.71,
|
| 919 |
+
"learning_rate": 9.61904761904762e-06,
|
| 920 |
+
"loss": 0.0517,
|
| 921 |
+
"step": 7600
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 1092.86,
|
| 925 |
+
"learning_rate": 9.023809523809524e-06,
|
| 926 |
+
"loss": 0.0527,
|
| 927 |
+
"step": 7650
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 1100.0,
|
| 931 |
+
"learning_rate": 8.428571428571429e-06,
|
| 932 |
+
"loss": 0.0533,
|
| 933 |
+
"step": 7700
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 1107.14,
|
| 937 |
+
"learning_rate": 7.833333333333333e-06,
|
| 938 |
+
"loss": 0.0537,
|
| 939 |
+
"step": 7750
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 1114.29,
|
| 943 |
+
"learning_rate": 7.238095238095238e-06,
|
| 944 |
+
"loss": 0.0517,
|
| 945 |
+
"step": 7800
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 1121.43,
|
| 949 |
+
"learning_rate": 6.642857142857144e-06,
|
| 950 |
+
"loss": 0.0517,
|
| 951 |
+
"step": 7850
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 1128.57,
|
| 955 |
+
"learning_rate": 6.047619047619048e-06,
|
| 956 |
+
"loss": 0.0506,
|
| 957 |
+
"step": 7900
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 1135.71,
|
| 961 |
+
"learning_rate": 5.452380952380952e-06,
|
| 962 |
+
"loss": 0.0512,
|
| 963 |
+
"step": 7950
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 1142.86,
|
| 967 |
+
"learning_rate": 4.857142857142858e-06,
|
| 968 |
+
"loss": 0.0482,
|
| 969 |
+
"step": 8000
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 1150.0,
|
| 973 |
+
"learning_rate": 4.261904761904762e-06,
|
| 974 |
+
"loss": 0.0488,
|
| 975 |
+
"step": 8050
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 1157.14,
|
| 979 |
+
"learning_rate": 3.666666666666667e-06,
|
| 980 |
+
"loss": 0.0493,
|
| 981 |
+
"step": 8100
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 1164.29,
|
| 985 |
+
"learning_rate": 3.0714285714285715e-06,
|
| 986 |
+
"loss": 0.0492,
|
| 987 |
+
"step": 8150
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 1171.43,
|
| 991 |
+
"learning_rate": 2.4761904761904764e-06,
|
| 992 |
+
"loss": 0.0491,
|
| 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": 4.803400934061458e+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 sha256:44c87eb99d8c175c5594d80edf27e113ccbd1489ec29e1e070d5088bf0ef55d5
|
| 3 |
+
size 4027
|