Upload 13 files
Browse files- config.json +114 -0
- merges.txt +0 -0
- optimizer.pt +3 -0
- preprocessor_config.json +27 -0
- pytorch_model.bin +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- special_tokens_map.json +51 -0
- tokenizer.json +0 -0
- tokenizer_config.json +87 -0
- trainer_state.json +676 -0
- training_args.bin +3 -0
- vocab.json +0 -0
config.json
ADDED
|
@@ -0,0 +1,114 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "sxandie/PickupReq_combined_RNS_Dec23_EEB_v4",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"LayoutLMv3ForTokenClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.1,
|
| 7 |
+
"bos_token_id": 0,
|
| 8 |
+
"classifier_dropout": null,
|
| 9 |
+
"coordinate_size": 128,
|
| 10 |
+
"eos_token_id": 2,
|
| 11 |
+
"has_relative_attention_bias": true,
|
| 12 |
+
"has_spatial_attention_bias": true,
|
| 13 |
+
"hidden_act": "gelu",
|
| 14 |
+
"hidden_dropout_prob": 0.1,
|
| 15 |
+
"hidden_size": 768,
|
| 16 |
+
"id2label": {
|
| 17 |
+
"0": "pick_up_date",
|
| 18 |
+
"1": "delivery_date",
|
| 19 |
+
"2": "client_reference",
|
| 20 |
+
"3": "weight_tot",
|
| 21 |
+
"4": "weight_measure_unit",
|
| 22 |
+
"5": "numpack_tot",
|
| 23 |
+
"6": "volumes_tot",
|
| 24 |
+
"7": "incoterms",
|
| 25 |
+
"8": "controlled_temp",
|
| 26 |
+
"9": "pickup_name",
|
| 27 |
+
"10": "pickup_department",
|
| 28 |
+
"11": "pickup_address",
|
| 29 |
+
"12": "pickup_city",
|
| 30 |
+
"13": "pickup_province",
|
| 31 |
+
"14": "pickup_zipcode",
|
| 32 |
+
"15": "pickup_state",
|
| 33 |
+
"16": "contact_email",
|
| 34 |
+
"17": "contact_phone",
|
| 35 |
+
"18": "delivery_name",
|
| 36 |
+
"19": "delivery_department",
|
| 37 |
+
"20": "delivery_address",
|
| 38 |
+
"21": "delivery_city",
|
| 39 |
+
"22": "delivery_province",
|
| 40 |
+
"23": "delivery_zipcode",
|
| 41 |
+
"24": "delivery_state",
|
| 42 |
+
"25": "item_numpack_body",
|
| 43 |
+
"26": "item_pack_type",
|
| 44 |
+
"27": "item_notes",
|
| 45 |
+
"28": "item_weight_body",
|
| 46 |
+
"29": "item_weight_body_measure_unit",
|
| 47 |
+
"30": "item_volume_body",
|
| 48 |
+
"31": "item_measures_body",
|
| 49 |
+
"32": "item_stackable",
|
| 50 |
+
"33": "goods_value",
|
| 51 |
+
"34": "others"
|
| 52 |
+
},
|
| 53 |
+
"initializer_range": 0.02,
|
| 54 |
+
"input_size": 224,
|
| 55 |
+
"intermediate_size": 3072,
|
| 56 |
+
"label2id": {
|
| 57 |
+
"client_reference": 2,
|
| 58 |
+
"contact_email": 16,
|
| 59 |
+
"contact_phone": 17,
|
| 60 |
+
"controlled_temp": 8,
|
| 61 |
+
"delivery_address": 20,
|
| 62 |
+
"delivery_city": 21,
|
| 63 |
+
"delivery_date": 1,
|
| 64 |
+
"delivery_department": 19,
|
| 65 |
+
"delivery_name": 18,
|
| 66 |
+
"delivery_province": 22,
|
| 67 |
+
"delivery_state": 24,
|
| 68 |
+
"delivery_zipcode": 23,
|
| 69 |
+
"goods_value": 33,
|
| 70 |
+
"incoterms": 7,
|
| 71 |
+
"item_measures_body": 31,
|
| 72 |
+
"item_notes": 27,
|
| 73 |
+
"item_numpack_body": 25,
|
| 74 |
+
"item_pack_type": 26,
|
| 75 |
+
"item_stackable": 32,
|
| 76 |
+
"item_volume_body": 30,
|
| 77 |
+
"item_weight_body": 28,
|
| 78 |
+
"item_weight_body_measure_unit": 29,
|
| 79 |
+
"numpack_tot": 5,
|
| 80 |
+
"others": 34,
|
| 81 |
+
"pick_up_date": 0,
|
| 82 |
+
"pickup_address": 11,
|
| 83 |
+
"pickup_city": 12,
|
| 84 |
+
"pickup_department": 10,
|
| 85 |
+
"pickup_name": 9,
|
| 86 |
+
"pickup_province": 13,
|
| 87 |
+
"pickup_state": 15,
|
| 88 |
+
"pickup_zipcode": 14,
|
| 89 |
+
"volumes_tot": 6,
|
| 90 |
+
"weight_measure_unit": 4,
|
| 91 |
+
"weight_tot": 3
|
| 92 |
+
},
|
| 93 |
+
"layer_norm_eps": 1e-05,
|
| 94 |
+
"max_2d_position_embeddings": 1024,
|
| 95 |
+
"max_position_embeddings": 514,
|
| 96 |
+
"max_rel_2d_pos": 256,
|
| 97 |
+
"max_rel_pos": 128,
|
| 98 |
+
"model_type": "layoutlmv3",
|
| 99 |
+
"num_attention_heads": 12,
|
| 100 |
+
"num_channels": 3,
|
| 101 |
+
"num_hidden_layers": 12,
|
| 102 |
+
"pad_token_id": 1,
|
| 103 |
+
"patch_size": 16,
|
| 104 |
+
"rel_2d_pos_bins": 64,
|
| 105 |
+
"rel_pos_bins": 32,
|
| 106 |
+
"second_input_size": 112,
|
| 107 |
+
"shape_size": 128,
|
| 108 |
+
"text_embed": true,
|
| 109 |
+
"torch_dtype": "float32",
|
| 110 |
+
"transformers_version": "4.28.0",
|
| 111 |
+
"type_vocab_size": 1,
|
| 112 |
+
"visual_embed": true,
|
| 113 |
+
"vocab_size": 50265
|
| 114 |
+
}
|
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b853f56d006fb347df53c0ba6ea514de902d3c757c0a9e5e2b5450c91c2f0bc9
|
| 3 |
+
size 1007683066
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"apply_ocr": false,
|
| 3 |
+
"do_normalize": true,
|
| 4 |
+
"do_rescale": true,
|
| 5 |
+
"do_resize": true,
|
| 6 |
+
"feature_extractor_type": "LayoutLMv3FeatureExtractor",
|
| 7 |
+
"image_mean": [
|
| 8 |
+
0.5,
|
| 9 |
+
0.5,
|
| 10 |
+
0.5
|
| 11 |
+
],
|
| 12 |
+
"image_processor_type": "LayoutLMv3ImageProcessor",
|
| 13 |
+
"image_std": [
|
| 14 |
+
0.5,
|
| 15 |
+
0.5,
|
| 16 |
+
0.5
|
| 17 |
+
],
|
| 18 |
+
"ocr_lang": null,
|
| 19 |
+
"processor_class": "LayoutLMv3Processor",
|
| 20 |
+
"resample": 2,
|
| 21 |
+
"rescale_factor": 0.00392156862745098,
|
| 22 |
+
"size": {
|
| 23 |
+
"height": 224,
|
| 24 |
+
"width": 224
|
| 25 |
+
},
|
| 26 |
+
"tesseract_config": ""
|
| 27 |
+
}
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2a6728cf455a1426ef990318d356fd48e1f350b311165de74454d6879d4f847e
|
| 3 |
+
size 503857390
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cc4069f95991f7fae62c25ecf3cf48b9debc51301c774598e699b6cb051540e5
|
| 3 |
+
size 14244
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3d6123b92a1a964482602c80eafadd952483e879e0467db7e659b5938ddcc1ab
|
| 3 |
+
size 1064
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "<s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": true,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"cls_token": {
|
| 10 |
+
"content": "<s>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": true,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"eos_token": {
|
| 17 |
+
"content": "</s>",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": true,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"mask_token": {
|
| 24 |
+
"content": "<mask>",
|
| 25 |
+
"lstrip": true,
|
| 26 |
+
"normalized": true,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
},
|
| 30 |
+
"pad_token": {
|
| 31 |
+
"content": "<pad>",
|
| 32 |
+
"lstrip": false,
|
| 33 |
+
"normalized": true,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false
|
| 36 |
+
},
|
| 37 |
+
"sep_token": {
|
| 38 |
+
"content": "</s>",
|
| 39 |
+
"lstrip": false,
|
| 40 |
+
"normalized": true,
|
| 41 |
+
"rstrip": false,
|
| 42 |
+
"single_word": false
|
| 43 |
+
},
|
| 44 |
+
"unk_token": {
|
| 45 |
+
"content": "<unk>",
|
| 46 |
+
"lstrip": false,
|
| 47 |
+
"normalized": true,
|
| 48 |
+
"rstrip": false,
|
| 49 |
+
"single_word": false
|
| 50 |
+
}
|
| 51 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,87 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": true,
|
| 3 |
+
"apply_ocr": false,
|
| 4 |
+
"bos_token": {
|
| 5 |
+
"__type": "AddedToken",
|
| 6 |
+
"content": "<s>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": true,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false
|
| 11 |
+
},
|
| 12 |
+
"clean_up_tokenization_spaces": true,
|
| 13 |
+
"cls_token": {
|
| 14 |
+
"__type": "AddedToken",
|
| 15 |
+
"content": "<s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": true,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false
|
| 20 |
+
},
|
| 21 |
+
"cls_token_box": [
|
| 22 |
+
0,
|
| 23 |
+
0,
|
| 24 |
+
0,
|
| 25 |
+
0
|
| 26 |
+
],
|
| 27 |
+
"eos_token": {
|
| 28 |
+
"__type": "AddedToken",
|
| 29 |
+
"content": "</s>",
|
| 30 |
+
"lstrip": false,
|
| 31 |
+
"normalized": true,
|
| 32 |
+
"rstrip": false,
|
| 33 |
+
"single_word": false
|
| 34 |
+
},
|
| 35 |
+
"errors": "replace",
|
| 36 |
+
"mask_token": {
|
| 37 |
+
"__type": "AddedToken",
|
| 38 |
+
"content": "<mask>",
|
| 39 |
+
"lstrip": true,
|
| 40 |
+
"normalized": true,
|
| 41 |
+
"rstrip": false,
|
| 42 |
+
"single_word": false
|
| 43 |
+
},
|
| 44 |
+
"model_max_length": 512,
|
| 45 |
+
"only_label_first_subword": true,
|
| 46 |
+
"pad_token": {
|
| 47 |
+
"__type": "AddedToken",
|
| 48 |
+
"content": "<pad>",
|
| 49 |
+
"lstrip": false,
|
| 50 |
+
"normalized": true,
|
| 51 |
+
"rstrip": false,
|
| 52 |
+
"single_word": false
|
| 53 |
+
},
|
| 54 |
+
"pad_token_box": [
|
| 55 |
+
0,
|
| 56 |
+
0,
|
| 57 |
+
0,
|
| 58 |
+
0
|
| 59 |
+
],
|
| 60 |
+
"pad_token_label": -100,
|
| 61 |
+
"processor_class": "LayoutLMv3Processor",
|
| 62 |
+
"sep_token": {
|
| 63 |
+
"__type": "AddedToken",
|
| 64 |
+
"content": "</s>",
|
| 65 |
+
"lstrip": false,
|
| 66 |
+
"normalized": true,
|
| 67 |
+
"rstrip": false,
|
| 68 |
+
"single_word": false
|
| 69 |
+
},
|
| 70 |
+
"sep_token_box": [
|
| 71 |
+
0,
|
| 72 |
+
0,
|
| 73 |
+
0,
|
| 74 |
+
0
|
| 75 |
+
],
|
| 76 |
+
"tokenizer_class": "LayoutLMv3Tokenizer",
|
| 77 |
+
"trim_offsets": true,
|
| 78 |
+
"trust_remote_code": false,
|
| 79 |
+
"unk_token": {
|
| 80 |
+
"__type": "AddedToken",
|
| 81 |
+
"content": "<unk>",
|
| 82 |
+
"lstrip": false,
|
| 83 |
+
"normalized": true,
|
| 84 |
+
"rstrip": false,
|
| 85 |
+
"single_word": false
|
| 86 |
+
}
|
| 87 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,676 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9824828636709825,
|
| 3 |
+
"best_model_checkpoint": "/content/drive/Shareddrives/Data Science Team folder-open/Clients/Rhenus Documents/attachments_batch7_20231027_SELECTED/Thermorossi/pdf2img/processedimg/0-60 THERMOROSSI/model/PickupReq_combined_RNS_Dec23_EEB_v4/checkpoint-4500",
|
| 4 |
+
"epoch": 78.125,
|
| 5 |
+
"global_step": 5000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.56,
|
| 12 |
+
"eval_accuracy": 0.9868951612903226,
|
| 13 |
+
"eval_f1": 0.942878903274943,
|
| 14 |
+
"eval_loss": 0.06447841972112656,
|
| 15 |
+
"eval_precision": 0.9435975609756098,
|
| 16 |
+
"eval_recall": 0.9421613394216134,
|
| 17 |
+
"eval_runtime": 1.2198,
|
| 18 |
+
"eval_samples_per_second": 10.658,
|
| 19 |
+
"eval_steps_per_second": 5.739,
|
| 20 |
+
"step": 100
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 3.12,
|
| 24 |
+
"eval_accuracy": 0.9914314516129032,
|
| 25 |
+
"eval_f1": 0.9573170731707317,
|
| 26 |
+
"eval_loss": 0.04372543469071388,
|
| 27 |
+
"eval_precision": 0.9587786259541985,
|
| 28 |
+
"eval_recall": 0.9558599695585996,
|
| 29 |
+
"eval_runtime": 1.2728,
|
| 30 |
+
"eval_samples_per_second": 10.213,
|
| 31 |
+
"eval_steps_per_second": 5.5,
|
| 32 |
+
"step": 200
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 4.69,
|
| 36 |
+
"eval_accuracy": 0.9924395161290323,
|
| 37 |
+
"eval_f1": 0.966360856269113,
|
| 38 |
+
"eval_loss": 0.04766829311847687,
|
| 39 |
+
"eval_precision": 0.9708141321044547,
|
| 40 |
+
"eval_recall": 0.9619482496194824,
|
| 41 |
+
"eval_runtime": 1.224,
|
| 42 |
+
"eval_samples_per_second": 10.621,
|
| 43 |
+
"eval_steps_per_second": 5.719,
|
| 44 |
+
"step": 300
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 6.25,
|
| 48 |
+
"eval_accuracy": 0.9939516129032258,
|
| 49 |
+
"eval_f1": 0.9786585365853657,
|
| 50 |
+
"eval_loss": 0.050641875714063644,
|
| 51 |
+
"eval_precision": 0.9801526717557252,
|
| 52 |
+
"eval_recall": 0.9771689497716894,
|
| 53 |
+
"eval_runtime": 1.2177,
|
| 54 |
+
"eval_samples_per_second": 10.676,
|
| 55 |
+
"eval_steps_per_second": 5.749,
|
| 56 |
+
"step": 400
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 7.81,
|
| 60 |
+
"learning_rate": 9e-06,
|
| 61 |
+
"loss": 0.145,
|
| 62 |
+
"step": 500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 7.81,
|
| 66 |
+
"eval_accuracy": 0.9939516129032258,
|
| 67 |
+
"eval_f1": 0.9733434881949734,
|
| 68 |
+
"eval_loss": 0.03310808911919594,
|
| 69 |
+
"eval_precision": 0.9740853658536586,
|
| 70 |
+
"eval_recall": 0.9726027397260274,
|
| 71 |
+
"eval_runtime": 1.2766,
|
| 72 |
+
"eval_samples_per_second": 10.183,
|
| 73 |
+
"eval_steps_per_second": 5.483,
|
| 74 |
+
"step": 500
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 9.38,
|
| 78 |
+
"eval_accuracy": 0.9939516129032258,
|
| 79 |
+
"eval_f1": 0.9718202589489718,
|
| 80 |
+
"eval_loss": 0.04493210092186928,
|
| 81 |
+
"eval_precision": 0.9725609756097561,
|
| 82 |
+
"eval_recall": 0.9710806697108066,
|
| 83 |
+
"eval_runtime": 1.2123,
|
| 84 |
+
"eval_samples_per_second": 10.724,
|
| 85 |
+
"eval_steps_per_second": 5.774,
|
| 86 |
+
"step": 600
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 10.94,
|
| 90 |
+
"eval_accuracy": 0.9939516129032258,
|
| 91 |
+
"eval_f1": 0.9702970297029704,
|
| 92 |
+
"eval_loss": 0.038082972168922424,
|
| 93 |
+
"eval_precision": 0.9710365853658537,
|
| 94 |
+
"eval_recall": 0.969558599695586,
|
| 95 |
+
"eval_runtime": 1.2843,
|
| 96 |
+
"eval_samples_per_second": 10.122,
|
| 97 |
+
"eval_steps_per_second": 5.45,
|
| 98 |
+
"step": 700
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 12.5,
|
| 102 |
+
"eval_accuracy": 0.9939516129032258,
|
| 103 |
+
"eval_f1": 0.9702970297029704,
|
| 104 |
+
"eval_loss": 0.03552929311990738,
|
| 105 |
+
"eval_precision": 0.9710365853658537,
|
| 106 |
+
"eval_recall": 0.969558599695586,
|
| 107 |
+
"eval_runtime": 1.2169,
|
| 108 |
+
"eval_samples_per_second": 10.683,
|
| 109 |
+
"eval_steps_per_second": 5.752,
|
| 110 |
+
"step": 800
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 14.06,
|
| 114 |
+
"eval_accuracy": 0.9919354838709677,
|
| 115 |
+
"eval_f1": 0.9672505712109674,
|
| 116 |
+
"eval_loss": 0.03628508001565933,
|
| 117 |
+
"eval_precision": 0.9679878048780488,
|
| 118 |
+
"eval_recall": 0.9665144596651446,
|
| 119 |
+
"eval_runtime": 1.2161,
|
| 120 |
+
"eval_samples_per_second": 10.69,
|
| 121 |
+
"eval_steps_per_second": 5.756,
|
| 122 |
+
"step": 900
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 15.62,
|
| 126 |
+
"learning_rate": 8.000000000000001e-06,
|
| 127 |
+
"loss": 0.0076,
|
| 128 |
+
"step": 1000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 15.62,
|
| 132 |
+
"eval_accuracy": 0.9919354838709677,
|
| 133 |
+
"eval_f1": 0.9642041127189643,
|
| 134 |
+
"eval_loss": 0.041032299399375916,
|
| 135 |
+
"eval_precision": 0.9649390243902439,
|
| 136 |
+
"eval_recall": 0.9634703196347032,
|
| 137 |
+
"eval_runtime": 1.2046,
|
| 138 |
+
"eval_samples_per_second": 10.792,
|
| 139 |
+
"eval_steps_per_second": 5.811,
|
| 140 |
+
"step": 1000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 17.19,
|
| 144 |
+
"eval_accuracy": 0.9939516129032258,
|
| 145 |
+
"eval_f1": 0.9763899466869764,
|
| 146 |
+
"eval_loss": 0.025884319096803665,
|
| 147 |
+
"eval_precision": 0.9771341463414634,
|
| 148 |
+
"eval_recall": 0.9756468797564688,
|
| 149 |
+
"eval_runtime": 1.2178,
|
| 150 |
+
"eval_samples_per_second": 10.675,
|
| 151 |
+
"eval_steps_per_second": 5.748,
|
| 152 |
+
"step": 1100
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 18.75,
|
| 156 |
+
"eval_accuracy": 0.9949596774193549,
|
| 157 |
+
"eval_f1": 0.9764258555133081,
|
| 158 |
+
"eval_loss": 0.03205409273505211,
|
| 159 |
+
"eval_precision": 0.9756838905775076,
|
| 160 |
+
"eval_recall": 0.9771689497716894,
|
| 161 |
+
"eval_runtime": 1.2104,
|
| 162 |
+
"eval_samples_per_second": 10.74,
|
| 163 |
+
"eval_steps_per_second": 5.783,
|
| 164 |
+
"step": 1200
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 20.31,
|
| 168 |
+
"eval_accuracy": 0.9949596774193549,
|
| 169 |
+
"eval_f1": 0.9764258555133081,
|
| 170 |
+
"eval_loss": 0.03990120440721512,
|
| 171 |
+
"eval_precision": 0.9756838905775076,
|
| 172 |
+
"eval_recall": 0.9771689497716894,
|
| 173 |
+
"eval_runtime": 1.2848,
|
| 174 |
+
"eval_samples_per_second": 10.118,
|
| 175 |
+
"eval_steps_per_second": 5.448,
|
| 176 |
+
"step": 1300
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 21.88,
|
| 180 |
+
"eval_accuracy": 0.9959677419354839,
|
| 181 |
+
"eval_f1": 0.9824828636709825,
|
| 182 |
+
"eval_loss": 0.026920977979898453,
|
| 183 |
+
"eval_precision": 0.9832317073170732,
|
| 184 |
+
"eval_recall": 0.9817351598173516,
|
| 185 |
+
"eval_runtime": 1.2238,
|
| 186 |
+
"eval_samples_per_second": 10.623,
|
| 187 |
+
"eval_steps_per_second": 5.72,
|
| 188 |
+
"step": 1400
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 23.44,
|
| 192 |
+
"learning_rate": 7e-06,
|
| 193 |
+
"loss": 0.0034,
|
| 194 |
+
"step": 1500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 23.44,
|
| 198 |
+
"eval_accuracy": 0.9944556451612904,
|
| 199 |
+
"eval_f1": 0.9763899466869764,
|
| 200 |
+
"eval_loss": 0.03660265728831291,
|
| 201 |
+
"eval_precision": 0.9771341463414634,
|
| 202 |
+
"eval_recall": 0.9756468797564688,
|
| 203 |
+
"eval_runtime": 1.2067,
|
| 204 |
+
"eval_samples_per_second": 10.773,
|
| 205 |
+
"eval_steps_per_second": 5.801,
|
| 206 |
+
"step": 1500
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 25.0,
|
| 210 |
+
"eval_accuracy": 0.9969758064516129,
|
| 211 |
+
"eval_f1": 0.9855293221629856,
|
| 212 |
+
"eval_loss": 0.0368938222527504,
|
| 213 |
+
"eval_precision": 0.9862804878048781,
|
| 214 |
+
"eval_recall": 0.984779299847793,
|
| 215 |
+
"eval_runtime": 1.2163,
|
| 216 |
+
"eval_samples_per_second": 10.688,
|
| 217 |
+
"eval_steps_per_second": 5.755,
|
| 218 |
+
"step": 1600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 26.56,
|
| 222 |
+
"eval_accuracy": 0.9954637096774194,
|
| 223 |
+
"eval_f1": 0.9794364051789796,
|
| 224 |
+
"eval_loss": 0.03973403945565224,
|
| 225 |
+
"eval_precision": 0.9801829268292683,
|
| 226 |
+
"eval_recall": 0.9786910197869102,
|
| 227 |
+
"eval_runtime": 1.2199,
|
| 228 |
+
"eval_samples_per_second": 10.657,
|
| 229 |
+
"eval_steps_per_second": 5.738,
|
| 230 |
+
"step": 1700
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 28.12,
|
| 234 |
+
"eval_accuracy": 0.9964717741935484,
|
| 235 |
+
"eval_f1": 0.9824828636709825,
|
| 236 |
+
"eval_loss": 0.027732405811548233,
|
| 237 |
+
"eval_precision": 0.9832317073170732,
|
| 238 |
+
"eval_recall": 0.9817351598173516,
|
| 239 |
+
"eval_runtime": 1.2739,
|
| 240 |
+
"eval_samples_per_second": 10.205,
|
| 241 |
+
"eval_steps_per_second": 5.495,
|
| 242 |
+
"step": 1800
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 29.69,
|
| 246 |
+
"eval_accuracy": 0.9964717741935484,
|
| 247 |
+
"eval_f1": 0.9855293221629856,
|
| 248 |
+
"eval_loss": 0.03053249605000019,
|
| 249 |
+
"eval_precision": 0.9862804878048781,
|
| 250 |
+
"eval_recall": 0.984779299847793,
|
| 251 |
+
"eval_runtime": 1.2239,
|
| 252 |
+
"eval_samples_per_second": 10.621,
|
| 253 |
+
"eval_steps_per_second": 5.719,
|
| 254 |
+
"step": 1900
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 31.25,
|
| 258 |
+
"learning_rate": 6e-06,
|
| 259 |
+
"loss": 0.0021,
|
| 260 |
+
"step": 2000
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 31.25,
|
| 264 |
+
"eval_accuracy": 0.9954637096774194,
|
| 265 |
+
"eval_f1": 0.9763899466869764,
|
| 266 |
+
"eval_loss": 0.04148135706782341,
|
| 267 |
+
"eval_precision": 0.9771341463414634,
|
| 268 |
+
"eval_recall": 0.9756468797564688,
|
| 269 |
+
"eval_runtime": 1.2182,
|
| 270 |
+
"eval_samples_per_second": 10.672,
|
| 271 |
+
"eval_steps_per_second": 5.746,
|
| 272 |
+
"step": 2000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 32.81,
|
| 276 |
+
"eval_accuracy": 0.9964717741935484,
|
| 277 |
+
"eval_f1": 0.9855293221629856,
|
| 278 |
+
"eval_loss": 0.03434097394347191,
|
| 279 |
+
"eval_precision": 0.9862804878048781,
|
| 280 |
+
"eval_recall": 0.984779299847793,
|
| 281 |
+
"eval_runtime": 1.2192,
|
| 282 |
+
"eval_samples_per_second": 10.663,
|
| 283 |
+
"eval_steps_per_second": 5.742,
|
| 284 |
+
"step": 2100
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 34.38,
|
| 288 |
+
"eval_accuracy": 0.9964717741935484,
|
| 289 |
+
"eval_f1": 0.9824828636709825,
|
| 290 |
+
"eval_loss": 0.036329954862594604,
|
| 291 |
+
"eval_precision": 0.9832317073170732,
|
| 292 |
+
"eval_recall": 0.9817351598173516,
|
| 293 |
+
"eval_runtime": 1.2759,
|
| 294 |
+
"eval_samples_per_second": 10.189,
|
| 295 |
+
"eval_steps_per_second": 5.486,
|
| 296 |
+
"step": 2200
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 35.94,
|
| 300 |
+
"eval_accuracy": 0.9959677419354839,
|
| 301 |
+
"eval_f1": 0.9824828636709825,
|
| 302 |
+
"eval_loss": 0.0368308424949646,
|
| 303 |
+
"eval_precision": 0.9832317073170732,
|
| 304 |
+
"eval_recall": 0.9817351598173516,
|
| 305 |
+
"eval_runtime": 1.215,
|
| 306 |
+
"eval_samples_per_second": 10.7,
|
| 307 |
+
"eval_steps_per_second": 5.762,
|
| 308 |
+
"step": 2300
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 37.5,
|
| 312 |
+
"eval_accuracy": 0.9964717741935484,
|
| 313 |
+
"eval_f1": 0.9824828636709825,
|
| 314 |
+
"eval_loss": 0.03690670430660248,
|
| 315 |
+
"eval_precision": 0.9832317073170732,
|
| 316 |
+
"eval_recall": 0.9817351598173516,
|
| 317 |
+
"eval_runtime": 1.2077,
|
| 318 |
+
"eval_samples_per_second": 10.764,
|
| 319 |
+
"eval_steps_per_second": 5.796,
|
| 320 |
+
"step": 2400
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 39.06,
|
| 324 |
+
"learning_rate": 5e-06,
|
| 325 |
+
"loss": 0.0016,
|
| 326 |
+
"step": 2500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 39.06,
|
| 330 |
+
"eval_accuracy": 0.9954637096774194,
|
| 331 |
+
"eval_f1": 0.9794364051789796,
|
| 332 |
+
"eval_loss": 0.03333241119980812,
|
| 333 |
+
"eval_precision": 0.9801829268292683,
|
| 334 |
+
"eval_recall": 0.9786910197869102,
|
| 335 |
+
"eval_runtime": 1.2147,
|
| 336 |
+
"eval_samples_per_second": 10.703,
|
| 337 |
+
"eval_steps_per_second": 5.763,
|
| 338 |
+
"step": 2500
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 40.62,
|
| 342 |
+
"eval_accuracy": 0.9979838709677419,
|
| 343 |
+
"eval_f1": 0.9916222391469918,
|
| 344 |
+
"eval_loss": 0.03542204946279526,
|
| 345 |
+
"eval_precision": 0.9923780487804879,
|
| 346 |
+
"eval_recall": 0.9908675799086758,
|
| 347 |
+
"eval_runtime": 1.2122,
|
| 348 |
+
"eval_samples_per_second": 10.725,
|
| 349 |
+
"eval_steps_per_second": 5.775,
|
| 350 |
+
"step": 2600
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 42.19,
|
| 354 |
+
"eval_accuracy": 0.9964717741935484,
|
| 355 |
+
"eval_f1": 0.9824828636709825,
|
| 356 |
+
"eval_loss": 0.03290965408086777,
|
| 357 |
+
"eval_precision": 0.9832317073170732,
|
| 358 |
+
"eval_recall": 0.9817351598173516,
|
| 359 |
+
"eval_runtime": 1.2189,
|
| 360 |
+
"eval_samples_per_second": 10.665,
|
| 361 |
+
"eval_steps_per_second": 5.743,
|
| 362 |
+
"step": 2700
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 43.75,
|
| 366 |
+
"eval_accuracy": 0.9969758064516129,
|
| 367 |
+
"eval_f1": 0.9855293221629856,
|
| 368 |
+
"eval_loss": 0.03801412135362625,
|
| 369 |
+
"eval_precision": 0.9862804878048781,
|
| 370 |
+
"eval_recall": 0.984779299847793,
|
| 371 |
+
"eval_runtime": 1.2022,
|
| 372 |
+
"eval_samples_per_second": 10.813,
|
| 373 |
+
"eval_steps_per_second": 5.823,
|
| 374 |
+
"step": 2800
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 45.31,
|
| 378 |
+
"eval_accuracy": 0.9959677419354839,
|
| 379 |
+
"eval_f1": 0.9794364051789796,
|
| 380 |
+
"eval_loss": 0.03737560659646988,
|
| 381 |
+
"eval_precision": 0.9801829268292683,
|
| 382 |
+
"eval_recall": 0.9786910197869102,
|
| 383 |
+
"eval_runtime": 1.2805,
|
| 384 |
+
"eval_samples_per_second": 10.152,
|
| 385 |
+
"eval_steps_per_second": 5.467,
|
| 386 |
+
"step": 2900
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 46.88,
|
| 390 |
+
"learning_rate": 4.000000000000001e-06,
|
| 391 |
+
"loss": 0.0008,
|
| 392 |
+
"step": 3000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 46.88,
|
| 396 |
+
"eval_accuracy": 0.9959677419354839,
|
| 397 |
+
"eval_f1": 0.9794364051789796,
|
| 398 |
+
"eval_loss": 0.03473510965704918,
|
| 399 |
+
"eval_precision": 0.9801829268292683,
|
| 400 |
+
"eval_recall": 0.9786910197869102,
|
| 401 |
+
"eval_runtime": 1.2112,
|
| 402 |
+
"eval_samples_per_second": 10.733,
|
| 403 |
+
"eval_steps_per_second": 5.779,
|
| 404 |
+
"step": 3000
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 48.44,
|
| 408 |
+
"eval_accuracy": 0.9959677419354839,
|
| 409 |
+
"eval_f1": 0.9794364051789796,
|
| 410 |
+
"eval_loss": 0.03780585899949074,
|
| 411 |
+
"eval_precision": 0.9801829268292683,
|
| 412 |
+
"eval_recall": 0.9786910197869102,
|
| 413 |
+
"eval_runtime": 1.2093,
|
| 414 |
+
"eval_samples_per_second": 10.75,
|
| 415 |
+
"eval_steps_per_second": 5.788,
|
| 416 |
+
"step": 3100
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 50.0,
|
| 420 |
+
"eval_accuracy": 0.9964717741935484,
|
| 421 |
+
"eval_f1": 0.9824828636709825,
|
| 422 |
+
"eval_loss": 0.03984608128666878,
|
| 423 |
+
"eval_precision": 0.9832317073170732,
|
| 424 |
+
"eval_recall": 0.9817351598173516,
|
| 425 |
+
"eval_runtime": 1.2798,
|
| 426 |
+
"eval_samples_per_second": 10.157,
|
| 427 |
+
"eval_steps_per_second": 5.469,
|
| 428 |
+
"step": 3200
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 51.56,
|
| 432 |
+
"eval_accuracy": 0.9959677419354839,
|
| 433 |
+
"eval_f1": 0.9794364051789796,
|
| 434 |
+
"eval_loss": 0.03601069748401642,
|
| 435 |
+
"eval_precision": 0.9801829268292683,
|
| 436 |
+
"eval_recall": 0.9786910197869102,
|
| 437 |
+
"eval_runtime": 1.2064,
|
| 438 |
+
"eval_samples_per_second": 10.776,
|
| 439 |
+
"eval_steps_per_second": 5.802,
|
| 440 |
+
"step": 3300
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 53.12,
|
| 444 |
+
"eval_accuracy": 0.9954637096774194,
|
| 445 |
+
"eval_f1": 0.9779131759329779,
|
| 446 |
+
"eval_loss": 0.04352593794465065,
|
| 447 |
+
"eval_precision": 0.9786585365853658,
|
| 448 |
+
"eval_recall": 0.9771689497716894,
|
| 449 |
+
"eval_runtime": 1.2834,
|
| 450 |
+
"eval_samples_per_second": 10.129,
|
| 451 |
+
"eval_steps_per_second": 5.454,
|
| 452 |
+
"step": 3400
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 54.69,
|
| 456 |
+
"learning_rate": 3e-06,
|
| 457 |
+
"loss": 0.0006,
|
| 458 |
+
"step": 3500
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 54.69,
|
| 462 |
+
"eval_accuracy": 0.9959677419354839,
|
| 463 |
+
"eval_f1": 0.9794364051789796,
|
| 464 |
+
"eval_loss": 0.044929005205631256,
|
| 465 |
+
"eval_precision": 0.9801829268292683,
|
| 466 |
+
"eval_recall": 0.9786910197869102,
|
| 467 |
+
"eval_runtime": 1.2158,
|
| 468 |
+
"eval_samples_per_second": 10.693,
|
| 469 |
+
"eval_steps_per_second": 5.758,
|
| 470 |
+
"step": 3500
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 56.25,
|
| 474 |
+
"eval_accuracy": 0.9959677419354839,
|
| 475 |
+
"eval_f1": 0.9794364051789796,
|
| 476 |
+
"eval_loss": 0.04337146133184433,
|
| 477 |
+
"eval_precision": 0.9801829268292683,
|
| 478 |
+
"eval_recall": 0.9786910197869102,
|
| 479 |
+
"eval_runtime": 1.3027,
|
| 480 |
+
"eval_samples_per_second": 9.979,
|
| 481 |
+
"eval_steps_per_second": 5.373,
|
| 482 |
+
"step": 3600
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 57.81,
|
| 486 |
+
"eval_accuracy": 0.9959677419354839,
|
| 487 |
+
"eval_f1": 0.9794364051789796,
|
| 488 |
+
"eval_loss": 0.042894065380096436,
|
| 489 |
+
"eval_precision": 0.9801829268292683,
|
| 490 |
+
"eval_recall": 0.9786910197869102,
|
| 491 |
+
"eval_runtime": 1.2493,
|
| 492 |
+
"eval_samples_per_second": 10.406,
|
| 493 |
+
"eval_steps_per_second": 5.603,
|
| 494 |
+
"step": 3700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 59.38,
|
| 498 |
+
"eval_accuracy": 0.9969758064516129,
|
| 499 |
+
"eval_f1": 0.9855293221629856,
|
| 500 |
+
"eval_loss": 0.04019169509410858,
|
| 501 |
+
"eval_precision": 0.9862804878048781,
|
| 502 |
+
"eval_recall": 0.984779299847793,
|
| 503 |
+
"eval_runtime": 1.2664,
|
| 504 |
+
"eval_samples_per_second": 10.265,
|
| 505 |
+
"eval_steps_per_second": 5.527,
|
| 506 |
+
"step": 3800
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 60.94,
|
| 510 |
+
"eval_accuracy": 0.9964717741935484,
|
| 511 |
+
"eval_f1": 0.9824828636709825,
|
| 512 |
+
"eval_loss": 0.041059188544750214,
|
| 513 |
+
"eval_precision": 0.9832317073170732,
|
| 514 |
+
"eval_recall": 0.9817351598173516,
|
| 515 |
+
"eval_runtime": 1.2339,
|
| 516 |
+
"eval_samples_per_second": 10.536,
|
| 517 |
+
"eval_steps_per_second": 5.673,
|
| 518 |
+
"step": 3900
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 62.5,
|
| 522 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 523 |
+
"loss": 0.0003,
|
| 524 |
+
"step": 4000
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 62.5,
|
| 528 |
+
"eval_accuracy": 0.9959677419354839,
|
| 529 |
+
"eval_f1": 0.9794364051789796,
|
| 530 |
+
"eval_loss": 0.04345503821969032,
|
| 531 |
+
"eval_precision": 0.9801829268292683,
|
| 532 |
+
"eval_recall": 0.9786910197869102,
|
| 533 |
+
"eval_runtime": 1.2758,
|
| 534 |
+
"eval_samples_per_second": 10.189,
|
| 535 |
+
"eval_steps_per_second": 5.487,
|
| 536 |
+
"step": 4000
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 64.06,
|
| 540 |
+
"eval_accuracy": 0.9959677419354839,
|
| 541 |
+
"eval_f1": 0.9794364051789796,
|
| 542 |
+
"eval_loss": 0.043661847710609436,
|
| 543 |
+
"eval_precision": 0.9801829268292683,
|
| 544 |
+
"eval_recall": 0.9786910197869102,
|
| 545 |
+
"eval_runtime": 1.2235,
|
| 546 |
+
"eval_samples_per_second": 10.625,
|
| 547 |
+
"eval_steps_per_second": 5.721,
|
| 548 |
+
"step": 4100
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 65.62,
|
| 552 |
+
"eval_accuracy": 0.9964717741935484,
|
| 553 |
+
"eval_f1": 0.9824828636709825,
|
| 554 |
+
"eval_loss": 0.04225226491689682,
|
| 555 |
+
"eval_precision": 0.9832317073170732,
|
| 556 |
+
"eval_recall": 0.9817351598173516,
|
| 557 |
+
"eval_runtime": 1.2314,
|
| 558 |
+
"eval_samples_per_second": 10.557,
|
| 559 |
+
"eval_steps_per_second": 5.685,
|
| 560 |
+
"step": 4200
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 67.19,
|
| 564 |
+
"eval_accuracy": 0.9969758064516129,
|
| 565 |
+
"eval_f1": 0.9855293221629856,
|
| 566 |
+
"eval_loss": 0.04165039211511612,
|
| 567 |
+
"eval_precision": 0.9862804878048781,
|
| 568 |
+
"eval_recall": 0.984779299847793,
|
| 569 |
+
"eval_runtime": 1.2205,
|
| 570 |
+
"eval_samples_per_second": 10.651,
|
| 571 |
+
"eval_steps_per_second": 5.735,
|
| 572 |
+
"step": 4300
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 68.75,
|
| 576 |
+
"eval_accuracy": 0.9964717741935484,
|
| 577 |
+
"eval_f1": 0.9824828636709825,
|
| 578 |
+
"eval_loss": 0.04333154484629631,
|
| 579 |
+
"eval_precision": 0.9832317073170732,
|
| 580 |
+
"eval_recall": 0.9817351598173516,
|
| 581 |
+
"eval_runtime": 1.2885,
|
| 582 |
+
"eval_samples_per_second": 10.089,
|
| 583 |
+
"eval_steps_per_second": 5.433,
|
| 584 |
+
"step": 4400
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 70.31,
|
| 588 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 589 |
+
"loss": 0.0003,
|
| 590 |
+
"step": 4500
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 70.31,
|
| 594 |
+
"eval_accuracy": 0.9964717741935484,
|
| 595 |
+
"eval_f1": 0.9824828636709825,
|
| 596 |
+
"eval_loss": 0.043314892798662186,
|
| 597 |
+
"eval_precision": 0.9832317073170732,
|
| 598 |
+
"eval_recall": 0.9817351598173516,
|
| 599 |
+
"eval_runtime": 1.2238,
|
| 600 |
+
"eval_samples_per_second": 10.623,
|
| 601 |
+
"eval_steps_per_second": 5.72,
|
| 602 |
+
"step": 4500
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 71.88,
|
| 606 |
+
"eval_accuracy": 0.9964717741935484,
|
| 607 |
+
"eval_f1": 0.9824828636709825,
|
| 608 |
+
"eval_loss": 0.043683238327503204,
|
| 609 |
+
"eval_precision": 0.9832317073170732,
|
| 610 |
+
"eval_recall": 0.9817351598173516,
|
| 611 |
+
"eval_runtime": 1.2463,
|
| 612 |
+
"eval_samples_per_second": 10.431,
|
| 613 |
+
"eval_steps_per_second": 5.616,
|
| 614 |
+
"step": 4600
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 73.44,
|
| 618 |
+
"eval_accuracy": 0.9969758064516129,
|
| 619 |
+
"eval_f1": 0.9855293221629856,
|
| 620 |
+
"eval_loss": 0.043529048562049866,
|
| 621 |
+
"eval_precision": 0.9862804878048781,
|
| 622 |
+
"eval_recall": 0.984779299847793,
|
| 623 |
+
"eval_runtime": 1.2409,
|
| 624 |
+
"eval_samples_per_second": 10.476,
|
| 625 |
+
"eval_steps_per_second": 5.641,
|
| 626 |
+
"step": 4700
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 75.0,
|
| 630 |
+
"eval_accuracy": 0.9964717741935484,
|
| 631 |
+
"eval_f1": 0.9824828636709825,
|
| 632 |
+
"eval_loss": 0.043888889253139496,
|
| 633 |
+
"eval_precision": 0.9832317073170732,
|
| 634 |
+
"eval_recall": 0.9817351598173516,
|
| 635 |
+
"eval_runtime": 1.2665,
|
| 636 |
+
"eval_samples_per_second": 10.265,
|
| 637 |
+
"eval_steps_per_second": 5.527,
|
| 638 |
+
"step": 4800
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 76.56,
|
| 642 |
+
"eval_accuracy": 0.9964717741935484,
|
| 643 |
+
"eval_f1": 0.9824828636709825,
|
| 644 |
+
"eval_loss": 0.04420410841703415,
|
| 645 |
+
"eval_precision": 0.9832317073170732,
|
| 646 |
+
"eval_recall": 0.9817351598173516,
|
| 647 |
+
"eval_runtime": 1.2292,
|
| 648 |
+
"eval_samples_per_second": 10.576,
|
| 649 |
+
"eval_steps_per_second": 5.695,
|
| 650 |
+
"step": 4900
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 78.12,
|
| 654 |
+
"learning_rate": 0.0,
|
| 655 |
+
"loss": 0.0002,
|
| 656 |
+
"step": 5000
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 78.12,
|
| 660 |
+
"eval_accuracy": 0.9964717741935484,
|
| 661 |
+
"eval_f1": 0.9824828636709825,
|
| 662 |
+
"eval_loss": 0.043921105563640594,
|
| 663 |
+
"eval_precision": 0.9832317073170732,
|
| 664 |
+
"eval_recall": 0.9817351598173516,
|
| 665 |
+
"eval_runtime": 1.2477,
|
| 666 |
+
"eval_samples_per_second": 10.419,
|
| 667 |
+
"eval_steps_per_second": 5.61,
|
| 668 |
+
"step": 5000
|
| 669 |
+
}
|
| 670 |
+
],
|
| 671 |
+
"max_steps": 5000,
|
| 672 |
+
"num_train_epochs": 79,
|
| 673 |
+
"total_flos": 2634149990123520.0,
|
| 674 |
+
"trial_name": null,
|
| 675 |
+
"trial_params": null
|
| 676 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:20a6bc362f6f50295438b22bad6d9c3653e9ab6d2ab8d6073d4e554a95b12abe
|
| 3 |
+
size 4408
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|