Training in progress, epoch 1
Browse files- .gitignore +1 -0
- config.json +195 -0
- pytorch_model.bin +3 -0
- runs/Aug23_10-09-21_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692785369.ip-172-31-42-230.eu-central-1.compute.internal.5099.0 +3 -0
- runs/Aug23_10-40-31_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692787239.ip-172-31-42-230.eu-central-1.compute.internal.5099.1 +3 -0
- runs/Aug23_11-42-10_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692790937.ip-172-31-42-230.eu-central-1.compute.internal.12495.0 +3 -0
- runs/Aug23_11-48-12_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692791376.ip-172-31-42-230.eu-central-1.compute.internal.12495.1 +3 -0
- runs/Aug23_11-57-19_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692792148.ip-172-31-42-230.eu-central-1.compute.internal.12495.2 +3 -0
- runs/Aug23_12-06-10_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692792457.ip-172-31-42-230.eu-central-1.compute.internal.12495.3 +3 -0
- runs/Aug23_12-10-20_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692792705.ip-172-31-42-230.eu-central-1.compute.internal.12495.4 +3 -0
- runs/Aug23_12-37-53_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692794438.ip-172-31-42-230.eu-central-1.compute.internal.12495.5 +3 -0
- runs/Aug23_13-15-45_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692796550.ip-172-31-42-230.eu-central-1.compute.internal.12495.6 +3 -0
- runs/Aug23_13-18-45_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692796811.ip-172-31-42-230.eu-central-1.compute.internal.12495.7 +3 -0
- runs/Aug23_13-36-38_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692797888.ip-172-31-42-230.eu-central-1.compute.internal.12495.8 +3 -0
- runs/Aug23_13-56-52_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692799016.ip-172-31-42-230.eu-central-1.compute.internal.4862.0 +3 -0
- training_args.bin +3 -0
.gitignore
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
checkpoint-*/
|
config.json
ADDED
@@ -0,0 +1,195 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_commit_hash": "8003d433113256b4ce3a0f5bf604b29ff78a7451",
|
3 |
+
"_name_or_path": "naver-clova-ix/donut-base-finetuned-cord-v2",
|
4 |
+
"architectures": [
|
5 |
+
"VisionEncoderDecoderModel"
|
6 |
+
],
|
7 |
+
"decoder": {
|
8 |
+
"_name_or_path": "",
|
9 |
+
"activation_dropout": 0.0,
|
10 |
+
"activation_function": "gelu",
|
11 |
+
"add_cross_attention": true,
|
12 |
+
"add_final_layer_norm": true,
|
13 |
+
"architectures": null,
|
14 |
+
"attention_dropout": 0.0,
|
15 |
+
"bad_words_ids": null,
|
16 |
+
"begin_suppress_tokens": null,
|
17 |
+
"bos_token_id": 0,
|
18 |
+
"chunk_size_feed_forward": 0,
|
19 |
+
"classifier_dropout": 0.0,
|
20 |
+
"cross_attention_hidden_size": null,
|
21 |
+
"d_model": 1024,
|
22 |
+
"decoder_attention_heads": 16,
|
23 |
+
"decoder_ffn_dim": 4096,
|
24 |
+
"decoder_layerdrop": 0.0,
|
25 |
+
"decoder_layers": 4,
|
26 |
+
"decoder_start_token_id": null,
|
27 |
+
"diversity_penalty": 0.0,
|
28 |
+
"do_sample": false,
|
29 |
+
"dropout": 0.1,
|
30 |
+
"early_stopping": false,
|
31 |
+
"encoder_attention_heads": 16,
|
32 |
+
"encoder_ffn_dim": 4096,
|
33 |
+
"encoder_layerdrop": 0.0,
|
34 |
+
"encoder_layers": 12,
|
35 |
+
"encoder_no_repeat_ngram_size": 0,
|
36 |
+
"eos_token_id": 2,
|
37 |
+
"exponential_decay_length_penalty": null,
|
38 |
+
"finetuning_task": null,
|
39 |
+
"forced_bos_token_id": null,
|
40 |
+
"forced_eos_token_id": 2,
|
41 |
+
"id2label": {
|
42 |
+
"0": "LABEL_0",
|
43 |
+
"1": "LABEL_1"
|
44 |
+
},
|
45 |
+
"init_std": 0.02,
|
46 |
+
"is_decoder": true,
|
47 |
+
"is_encoder_decoder": false,
|
48 |
+
"label2id": {
|
49 |
+
"LABEL_0": 0,
|
50 |
+
"LABEL_1": 1
|
51 |
+
},
|
52 |
+
"length_penalty": 1.0,
|
53 |
+
"max_length": 512,
|
54 |
+
"max_position_embeddings": 768,
|
55 |
+
"min_length": 0,
|
56 |
+
"model_type": "mbart",
|
57 |
+
"no_repeat_ngram_size": 0,
|
58 |
+
"num_beam_groups": 1,
|
59 |
+
"num_beams": 1,
|
60 |
+
"num_hidden_layers": 12,
|
61 |
+
"num_return_sequences": 1,
|
62 |
+
"output_attentions": false,
|
63 |
+
"output_hidden_states": false,
|
64 |
+
"output_scores": false,
|
65 |
+
"pad_token_id": 1,
|
66 |
+
"prefix": null,
|
67 |
+
"problem_type": null,
|
68 |
+
"pruned_heads": {},
|
69 |
+
"remove_invalid_values": false,
|
70 |
+
"repetition_penalty": 1.0,
|
71 |
+
"return_dict": true,
|
72 |
+
"return_dict_in_generate": false,
|
73 |
+
"scale_embedding": true,
|
74 |
+
"sep_token_id": null,
|
75 |
+
"suppress_tokens": null,
|
76 |
+
"task_specific_params": null,
|
77 |
+
"temperature": 1.0,
|
78 |
+
"tf_legacy_loss": false,
|
79 |
+
"tie_encoder_decoder": false,
|
80 |
+
"tie_word_embeddings": true,
|
81 |
+
"tokenizer_class": null,
|
82 |
+
"top_k": 50,
|
83 |
+
"top_p": 1.0,
|
84 |
+
"torch_dtype": null,
|
85 |
+
"torchscript": false,
|
86 |
+
"transformers_version": "4.30.1",
|
87 |
+
"typical_p": 1.0,
|
88 |
+
"use_bfloat16": false,
|
89 |
+
"use_cache": true,
|
90 |
+
"vocab_size": 57580
|
91 |
+
},
|
92 |
+
"decoder_start_token_id": 0,
|
93 |
+
"encoder": {
|
94 |
+
"_name_or_path": "",
|
95 |
+
"add_cross_attention": false,
|
96 |
+
"architectures": null,
|
97 |
+
"attention_probs_dropout_prob": 0.0,
|
98 |
+
"bad_words_ids": null,
|
99 |
+
"begin_suppress_tokens": null,
|
100 |
+
"bos_token_id": null,
|
101 |
+
"chunk_size_feed_forward": 0,
|
102 |
+
"cross_attention_hidden_size": null,
|
103 |
+
"decoder_start_token_id": null,
|
104 |
+
"depths": [
|
105 |
+
2,
|
106 |
+
2,
|
107 |
+
14,
|
108 |
+
2
|
109 |
+
],
|
110 |
+
"diversity_penalty": 0.0,
|
111 |
+
"do_sample": false,
|
112 |
+
"drop_path_rate": 0.1,
|
113 |
+
"early_stopping": false,
|
114 |
+
"embed_dim": 128,
|
115 |
+
"encoder_no_repeat_ngram_size": 0,
|
116 |
+
"eos_token_id": null,
|
117 |
+
"exponential_decay_length_penalty": null,
|
118 |
+
"finetuning_task": null,
|
119 |
+
"forced_bos_token_id": null,
|
120 |
+
"forced_eos_token_id": null,
|
121 |
+
"hidden_act": "gelu",
|
122 |
+
"hidden_dropout_prob": 0.0,
|
123 |
+
"hidden_size": 1024,
|
124 |
+
"id2label": {
|
125 |
+
"0": "LABEL_0",
|
126 |
+
"1": "LABEL_1"
|
127 |
+
},
|
128 |
+
"image_size": [
|
129 |
+
960,
|
130 |
+
720
|
131 |
+
],
|
132 |
+
"initializer_range": 0.02,
|
133 |
+
"is_decoder": false,
|
134 |
+
"is_encoder_decoder": false,
|
135 |
+
"label2id": {
|
136 |
+
"LABEL_0": 0,
|
137 |
+
"LABEL_1": 1
|
138 |
+
},
|
139 |
+
"layer_norm_eps": 1e-05,
|
140 |
+
"length_penalty": 1.0,
|
141 |
+
"max_length": 20,
|
142 |
+
"min_length": 0,
|
143 |
+
"mlp_ratio": 4.0,
|
144 |
+
"model_type": "donut-swin",
|
145 |
+
"no_repeat_ngram_size": 0,
|
146 |
+
"num_beam_groups": 1,
|
147 |
+
"num_beams": 1,
|
148 |
+
"num_channels": 3,
|
149 |
+
"num_heads": [
|
150 |
+
4,
|
151 |
+
8,
|
152 |
+
16,
|
153 |
+
32
|
154 |
+
],
|
155 |
+
"num_layers": 4,
|
156 |
+
"num_return_sequences": 1,
|
157 |
+
"output_attentions": false,
|
158 |
+
"output_hidden_states": false,
|
159 |
+
"output_scores": false,
|
160 |
+
"pad_token_id": null,
|
161 |
+
"patch_size": 4,
|
162 |
+
"path_norm": true,
|
163 |
+
"prefix": null,
|
164 |
+
"problem_type": null,
|
165 |
+
"pruned_heads": {},
|
166 |
+
"qkv_bias": true,
|
167 |
+
"remove_invalid_values": false,
|
168 |
+
"repetition_penalty": 1.0,
|
169 |
+
"return_dict": true,
|
170 |
+
"return_dict_in_generate": false,
|
171 |
+
"sep_token_id": null,
|
172 |
+
"suppress_tokens": null,
|
173 |
+
"task_specific_params": null,
|
174 |
+
"temperature": 1.0,
|
175 |
+
"tf_legacy_loss": false,
|
176 |
+
"tie_encoder_decoder": false,
|
177 |
+
"tie_word_embeddings": true,
|
178 |
+
"tokenizer_class": null,
|
179 |
+
"top_k": 50,
|
180 |
+
"top_p": 1.0,
|
181 |
+
"torch_dtype": null,
|
182 |
+
"torchscript": false,
|
183 |
+
"transformers_version": "4.30.1",
|
184 |
+
"typical_p": 1.0,
|
185 |
+
"use_absolute_embeddings": false,
|
186 |
+
"use_bfloat16": false,
|
187 |
+
"window_size": 10
|
188 |
+
},
|
189 |
+
"is_encoder_decoder": true,
|
190 |
+
"model_type": "vision-encoder-decoder",
|
191 |
+
"pad_token_id": 1,
|
192 |
+
"tie_word_embeddings": false,
|
193 |
+
"torch_dtype": "float32",
|
194 |
+
"transformers_version": null
|
195 |
+
}
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:700ed2d59a81d80f546a6e716fe1c96a7e1ae1b1cdd0695ea941af455ddc099f
|
3 |
+
size 806258457
|
runs/Aug23_10-09-21_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692785369.ip-172-31-42-230.eu-central-1.compute.internal.5099.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e5be2aefc09bb8b8ec5cccb270414a6eb8d10f778d1c86b5284a0aaca3030ca7
|
3 |
+
size 8553
|
runs/Aug23_10-40-31_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692787239.ip-172-31-42-230.eu-central-1.compute.internal.5099.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:001c242faa09262c4dd51da32375494758d830e0e2200b3ad12b14d2232d7993
|
3 |
+
size 8553
|
runs/Aug23_11-42-10_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692790937.ip-172-31-42-230.eu-central-1.compute.internal.12495.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:532a34488d25cc4a8d6af55ce1d76b93ca99ef98d912b8093980df1a2d54a0d1
|
3 |
+
size 8553
|
runs/Aug23_11-48-12_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692791376.ip-172-31-42-230.eu-central-1.compute.internal.12495.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:67aa73af7278de65eb2364c292b0b964bcfcb7f972f3be86f69a51e639a0c62c
|
3 |
+
size 8553
|
runs/Aug23_11-57-19_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692792148.ip-172-31-42-230.eu-central-1.compute.internal.12495.2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b5ddfe2141039d30d675b48a6e58ebbe48533c71af31055d621a1028b4fd3a6c
|
3 |
+
size 8553
|
runs/Aug23_12-06-10_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692792457.ip-172-31-42-230.eu-central-1.compute.internal.12495.3
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2b2e6f7becbabe1d807c3a0675464a2a72c732ddfceedef1e36f6276d1a81b6c
|
3 |
+
size 8553
|
runs/Aug23_12-10-20_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692792705.ip-172-31-42-230.eu-central-1.compute.internal.12495.4
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:82f7343080560be3a668d7884e7a800cdec5c7a5b575fb97cfee1250d4ee8645
|
3 |
+
size 8536
|
runs/Aug23_12-37-53_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692794438.ip-172-31-42-230.eu-central-1.compute.internal.12495.5
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cf73688973399d927dac0f5df2a0c4877fe0253ad795206bdf33f31fae3a9345
|
3 |
+
size 8554
|
runs/Aug23_13-15-45_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692796550.ip-172-31-42-230.eu-central-1.compute.internal.12495.6
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:caa81d9ca8d0b4f9e679205535dbe0c1eb7fe94a3685018e120e680c93ed18f6
|
3 |
+
size 8554
|
runs/Aug23_13-18-45_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692796811.ip-172-31-42-230.eu-central-1.compute.internal.12495.7
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7508cd047db3524221dde34b39cb54ef22f2e6902869499407010b3160f7c901
|
3 |
+
size 8554
|
runs/Aug23_13-36-38_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692797888.ip-172-31-42-230.eu-central-1.compute.internal.12495.8
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7910895f31e515e118d11b18f7217328a93a4f6501d6d83f9200d146fbe326f5
|
3 |
+
size 8554
|
runs/Aug23_13-56-52_ip-172-31-42-230.eu-central-1.compute.internal/events.out.tfevents.1692799016.ip-172-31-42-230.eu-central-1.compute.internal.4862.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3ae1d67516334dac386646702286219196b8adbe85d69cdaa1d2f8af32bbb2ba
|
3 |
+
size 8708
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7bbd4ef8804b9cc7533ca3dd90ece2bf54b7819d86f70a1ced20e1bb312031c0
|
3 |
+
size 4155
|