Hugging Face's logo

zuzannad1
/
pixelsum_medium

Transformers
PyTorch
pixelsum
Inference Endpoints
Model card Files Community
zuzannad1 commited on
Commit
629cef5
•
1 Parent(s): 7b8e74a

Upload 7 files

Browse files
Files changed (7) hide show
  1. config.json +191 -0
  2. merges.txt +0 -0
  3. pytorch_model.bin +3 -0
  4. special_tokens_map.json +1 -0
  5. text_renderer_config.json +232 -0
  6. tokenizer.json +0 -0
  7. tokenizer_config.json +1 -0
config.json ADDED
@@ -0,0 +1,191 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "PIXELSumModel"
4
+ ],
5
+ "decoder": {
6
+ "_name_or_path": "gpt2-medium",
7
+ "activation_function": "gelu_new",
8
+ "add_cross_attention": true,
9
+ "architectures": [
10
+ "GPT2LMHeadModel"
11
+ ],
12
+ "attn_pdrop": 0.1,
13
+ "bad_words_ids": null,
14
+ "bos_token_id": 50256,
15
+ "chunk_size_feed_forward": 0,
16
+ "cross_attention_hidden_size": null,
17
+ "cross_attention_reduce_factor": 1,
18
+ "decoder_start_token_id": null,
19
+ "diversity_penalty": 0.0,
20
+ "do_sample": false,
21
+ "early_stopping": false,
22
+ "embd_pdrop": 0.1,
23
+ "encoder_hidden_size": 768,
24
+ "encoder_no_repeat_ngram_size": 0,
25
+ "eos_token_id": 50256,
26
+ "finetuning_task": null,
27
+ "forced_bos_token_id": null,
28
+ "forced_eos_token_id": null,
29
+ "id2label": {
30
+ "0": "LABEL_0",
31
+ "1": "LABEL_1"
32
+ },
33
+ "initializer_range": 0.02,
34
+ "is_decoder": true,
35
+ "is_encoder_decoder": false,
36
+ "label2id": {
37
+ "LABEL_0": 0,
38
+ "LABEL_1": 1
39
+ },
40
+ "layer_norm_epsilon": 1e-05,
41
+ "length_penalty": 1.0,
42
+ "max_length": 20,
43
+ "min_length": 0,
44
+ "model_type": "gpt2",
45
+ "n_ctx": 1024,
46
+ "n_embd": 1024,
47
+ "n_head": 16,
48
+ "n_inner": null,
49
+ "n_layer": 24,
50
+ "n_positions": 1024,
51
+ "n_special": 0,
52
+ "no_repeat_ngram_size": 0,
53
+ "num_beam_groups": 1,
54
+ "num_beams": 1,
55
+ "num_return_sequences": 1,
56
+ "output_attentions": false,
57
+ "output_hidden_states": false,
58
+ "output_scores": false,
59
+ "pad_token_id": null,
60
+ "predict_special_tokens": true,
61
+ "prefix": null,
62
+ "problem_type": null,
63
+ "pruned_heads": {},
64
+ "remove_invalid_values": false,
65
+ "reorder_and_upcast_attn": false,
66
+ "repetition_penalty": 1.0,
67
+ "resid_pdrop": 0.1,
68
+ "return_dict": true,
69
+ "return_dict_in_generate": false,
70
+ "scale_attn_by_inverse_layer_idx": false,
71
+ "scale_attn_weights": true,
72
+ "sep_token_id": null,
73
+ "summary_activation": null,
74
+ "summary_first_dropout": 0.1,
75
+ "summary_proj_to_labels": true,
76
+ "summary_type": "cls_index",
77
+ "summary_use_proj": true,
78
+ "task_specific_params": {
79
+ "text-generation": {
80
+ "do_sample": true,
81
+ "max_length": 50
82
+ }
83
+ },
84
+ "temperature": 1.0,
85
+ "tie_encoder_decoder": false,
86
+ "tie_word_embeddings": true,
87
+ "tokenizer_class": null,
88
+ "top_k": 50,
89
+ "top_p": 1.0,
90
+ "torch_dtype": null,
91
+ "torchscript": false,
92
+ "transformers_version": "4.17.0",
93
+ "typical_p": 1.0,
94
+ "use_bfloat16": false,
95
+ "use_cache": true,
96
+ "vocab_size": 50257
97
+ },
98
+ "decoder_start_token_id": 50256,
99
+ "encoder": {
100
+ "_name_or_path": "Team-PIXEL/pixel-base",
101
+ "add_cross_attention": false,
102
+ "architectures": [
103
+ "PIXELForPreTraining"
104
+ ],
105
+ "attention_probs_dropout_prob": 0.1,
106
+ "bad_words_ids": null,
107
+ "bos_token_id": null,
108
+ "chunk_size_feed_forward": 0,
109
+ "cross_attention_hidden_size": null,
110
+ "decoder_hidden_size": 512,
111
+ "decoder_intermediate_size": 2048,
112
+ "decoder_num_attention_heads": 16,
113
+ "decoder_num_hidden_layers": 8,
114
+ "decoder_start_token_id": null,
115
+ "diversity_penalty": 0.0,
116
+ "do_sample": false,
117
+ "early_stopping": false,
118
+ "encoder_no_repeat_ngram_size": 0,
119
+ "eos_token_id": null,
120
+ "finetuning_task": null,
121
+ "forced_bos_token_id": null,
122
+ "forced_eos_token_id": null,
123
+ "hidden_act": "gelu",
124
+ "hidden_dropout_prob": 0.1,
125
+ "hidden_size": 768,
126
+ "id2label": {
127
+ "0": "LABEL_0",
128
+ "1": "LABEL_1"
129
+ },
130
+ "image_size": [
131
+ 16,
132
+ 8464
133
+ ],
134
+ "initializer_range": 0.02,
135
+ "intermediate_size": 3072,
136
+ "is_decoder": false,
137
+ "is_encoder_decoder": false,
138
+ "label2id": {
139
+ "LABEL_0": 0,
140
+ "LABEL_1": 1
141
+ },
142
+ "layer_norm_eps": 1e-12,
143
+ "length_penalty": 1.0,
144
+ "mask_ratio": 0.25,
145
+ "max_length": 20,
146
+ "min_length": 0,
147
+ "model_type": "pixel",
148
+ "no_repeat_ngram_size": 0,
149
+ "norm_pix_loss": true,
150
+ "num_attention_heads": 12,
151
+ "num_beam_groups": 1,
152
+ "num_beams": 1,
153
+ "num_channels": 3,
154
+ "num_hidden_layers": 12,
155
+ "num_return_sequences": 1,
156
+ "output_attentions": false,
157
+ "output_hidden_states": false,
158
+ "output_scores": false,
159
+ "pad_token_id": null,
160
+ "patch_size": 16,
161
+ "prefix": null,
162
+ "problem_type": null,
163
+ "pruned_heads": {},
164
+ "qkv_bias": true,
165
+ "remove_invalid_values": false,
166
+ "repetition_penalty": 1.0,
167
+ "return_dict": true,
168
+ "return_dict_in_generate": false,
169
+ "sep_token_id": null,
170
+ "task_specific_params": null,
171
+ "temperature": 1.0,
172
+ "tie_encoder_decoder": false,
173
+ "tie_word_embeddings": true,
174
+ "tokenizer_class": null,
175
+ "top_k": 50,
176
+ "top_p": 1.0,
177
+ "torch_dtype": "float32",
178
+ "torchscript": false,
179
+ "transformers_version": "4.17.0",
180
+ "typical_p": 1.0,
181
+ "use_bfloat16": false
182
+ },
183
+ "eos_token_id": 50256,
184
+ "is_encoder_decoder": true,
185
+ "model_type": "pixelsum",
186
+ "pad_token_id": 50256,
187
+ "tie_word_embeddings": false,
188
+ "torch_dtype": "float32",
189
+ "transformers_version": null,
190
+ "vocab_size": 50257
191
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:227f8688bcf1d5e6f678de9325bf00d7c6dda32c1068ddd0d8d9a35445787d91
3
+ size 2220545213
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"bos_token": "<|endoftext|>", "eos_token": "<|endoftext|>", "unk_token": "<|endoftext|>", "pad_token": "<|pad|>"}
text_renderer_config.json ADDED
@@ -0,0 +1,232 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "background_color": "white",
3
+ "dpi": 120,
4
+ "font_color": "black",
5
+ "font_file": "49e6dc219d1a1a1c9236acaf05a48b542002016a6dc877ee72baab085a84257b.3f28e7f4b38e1efe1b6da4a3732404c19d4c6a614ff32dce90a251e293d4ce58",
6
+ "font_size": 8,
7
+ "fonts_list": [
8
+ "Apple Color Emoji",
9
+ "C059",
10
+ "D050000L",
11
+ "DejaVu Sans",
12
+ "Go Noto Current",
13
+ "Inconsolata",
14
+ "Monospace",
15
+ "Nimbus Mono PS",
16
+ "Nimbus Roman",
17
+ "Nimbus Sans",
18
+ "Nimbus Sans Narrow",
19
+ "Noto Fangsong KSS Rotated",
20
+ "Noto Fangsong KSS Vertical",
21
+ "Noto Kufi Arabic",
22
+ "Noto Music",
23
+ "Noto Naskh Arabic",
24
+ "Noto Naskh Arabic UI",
25
+ "Noto Nastaliq Urdu",
26
+ "Noto Rashi Hebrew",
27
+ "Noto Sans",
28
+ "Noto Sans Adlam",
29
+ "Noto Sans Adlam Unjoined",
30
+ "Noto Sans Anatolian Hieroglyphs",
31
+ "Noto Sans Arabic",
32
+ "Noto Sans Arabic UI",
33
+ "Noto Sans Armenian",
34
+ "Noto Sans Avestan",
35
+ "Noto Sans Balinese",
36
+ "Noto Sans Bamum",
37
+ "Noto Sans Bassa Vah",
38
+ "Noto Sans Batak",
39
+ "Noto Sans Bengali",
40
+ "Noto Sans Bengali UI",
41
+ "Noto Sans Bhaiksuki",
42
+ "Noto Sans Brahmi",
43
+ "Noto Sans Buginese",
44
+ "Noto Sans Buhid",
45
+ "Noto Sans CJK HK",
46
+ "Noto Sans CJK JP",
47
+ "Noto Sans CJK KR",
48
+ "Noto Sans CJK SC",
49
+ "Noto Sans CJK TC",
50
+ "Noto Sans Canadian Aboriginal",
51
+ "Noto Sans Carian",
52
+ "Noto Sans Caucasian Albanian",
53
+ "Noto Sans Chakma",
54
+ "Noto Sans Cham",
55
+ "Noto Sans Cherokee",
56
+ "Noto Sans Chorasmian",
57
+ "Noto Sans Coptic",
58
+ "Noto Sans Cuneiform",
59
+ "Noto Sans Cypriot",
60
+ "Noto Sans Cypro Minoan",
61
+ "Noto Sans Deseret",
62
+ "Noto Sans Devanagari",
63
+ "Noto Sans Devanagari UI",
64
+ "Noto Sans Duployan",
65
+ "Noto Sans Egyptian Hieroglyphs",
66
+ "Noto Sans Elbasan",
67
+ "Noto Sans Elymaic",
68
+ "Noto Sans Ethiopic",
69
+ "Noto Sans Georgian",
70
+ "Noto Sans Glagolitic",
71
+ "Noto Sans Gothic",
72
+ "Noto Sans Grantha",
73
+ "Noto Sans Gujarati",
74
+ "Noto Sans Gujarati UI",
75
+ "Noto Sans Gunjala Gondi",
76
+ "Noto Sans Gurmukhi",
77
+ "Noto Sans Gurmukhi UI",
78
+ "Noto Sans Hanifi Rohingya",
79
+ "Noto Sans Hanunoo",
80
+ "Noto Sans Hatran",
81
+ "Noto Sans Hebrew",
82
+ "Noto Sans Hebrew Droid",
83
+ "Noto Sans Imperial Aramaic",
84
+ "Noto Sans Indic Siyaq Numbers",
85
+ "Noto Sans Inscriptional Pahlavi",
86
+ "Noto Sans Inscriptional Parthian",
87
+ "Noto Sans Javanese",
88
+ "Noto Sans Kaithi",
89
+ "Noto Sans Kannada",
90
+ "Noto Sans Kannada UI",
91
+ "Noto Sans Kawi",
92
+ "Noto Sans Kayah Li",
93
+ "Noto Sans Kharoshthi",
94
+ "Noto Sans Khmer",
95
+ "Noto Sans Khmer UI",
96
+ "Noto Sans Khojki",
97
+ "Noto Sans Khudawadi",
98
+ "Noto Sans Lao",
99
+ "Noto Sans Lao Looped",
100
+ "Noto Sans Lao Looped UI",
101
+ "Noto Sans Lao UI",
102
+ "Noto Sans Lepcha",
103
+ "Noto Sans Limbu",
104
+ "Noto Sans Linear A",
105
+ "Noto Sans Linear B",
106
+ "Noto Sans Lisu",
107
+ "Noto Sans Lycian",
108
+ "Noto Sans Lydian",
109
+ "Noto Sans Mahajani",
110
+ "Noto Sans Malayalam",
111
+ "Noto Sans Malayalam UI",
112
+ "Noto Sans Mandaic",
113
+ "Noto Sans Manichaean",
114
+ "Noto Sans Marchen",
115
+ "Noto Sans Masaram Gondi",
116
+ "Noto Sans Mayan Numerals",
117
+ "Noto Sans Medefaidrin",
118
+ "Noto Sans Meetei Mayek",
119
+ "Noto Sans Mende Kikakui",
120
+ "Noto Sans Meroitic",
121
+ "Noto Sans Miao",
122
+ "Noto Sans Modi",
123
+ "Noto Sans Mongolian",
124
+ "Noto Sans Mono",
125
+ "Noto Sans Mro",
126
+ "Noto Sans Multani",
127
+ "Noto Sans Myanmar",
128
+ "Noto Sans Myanmar UI",
129
+ "Noto Sans NKo",
130
+ "Noto Sans NKo Unjoined",
131
+ "Noto Sans Nabataean",
132
+ "Noto Sans Nag Mundari",
133
+ "Noto Sans Nandinagari",
134
+ "Noto Sans New Tai Lue",
135
+ "Noto Sans Newa",
136
+ "Noto Sans Nushu",
137
+ "Noto Sans Ogham",
138
+ "Noto Sans Ol Chiki",
139
+ "Noto Sans Old Hungarian",
140
+ "Noto Sans Old Italic",
141
+ "Noto Sans Old North Arabian",
142
+ "Noto Sans Old Permic",
143
+ "Noto Sans Old Persian",
144
+ "Noto Sans Old Sogdian",
145
+ "Noto Sans Old South Arabian",
146
+ "Noto Sans Old Turkic",
147
+ "Noto Sans Oriya",
148
+ "Noto Sans Osage",
149
+ "Noto Sans Osmanya",
150
+ "Noto Sans Pahawh Hmong",
151
+ "Noto Sans Palmyrene",
152
+ "Noto Sans Pau Cin Hau",
153
+ "Noto Sans Phags-Pa",
154
+ "Noto Sans Phoenician",
155
+ "Noto Sans Psalter Pahlavi",
156
+ "Noto Sans Rejang",
157
+ "Noto Sans Runic",
158
+ "Noto Sans Samaritan",
159
+ "Noto Sans Saurashtra",
160
+ "Noto Sans Sharada",
161
+ "Noto Sans Shavian",
162
+ "Noto Sans Siddham",
163
+ "Noto Sans SignWriting",
164
+ "Noto Sans Sinhala",
165
+ "Noto Sans Sinhala UI",
166
+ "Noto Sans Sogdian",
167
+ "Noto Sans Sora Sompeng",
168
+ "Noto Sans Soyombo",
169
+ "Noto Sans Sundanese",
170
+ "Noto Sans Syloti Nagri",
171
+ "Noto Sans Symbols",
172
+ "Noto Sans Symbols 2",
173
+ "Noto Sans Syriac",
174
+ "Noto Sans Syriac Eastern",
175
+ "Noto Sans Syriac Western",
176
+ "Noto Sans Tagalog",
177
+ "Noto Sans Tagbanwa",
178
+ "Noto Sans Tai Le",
179
+ "Noto Sans Tai Tham",
180
+ "Noto Sans Tai Viet",
181
+ "Noto Sans Takri",
182
+ "Noto Sans Tamil",
183
+ "Noto Sans Tamil Supplement",
184
+ "Noto Sans Tamil UI",
185
+ "Noto Sans Tangsa",
186
+ "Noto Sans Telugu",
187
+ "Noto Sans Telugu UI",
188
+ "Noto Sans Test",
189
+ "Noto Sans Thaana",
190
+ "Noto Sans Thai",
191
+ "Noto Sans Thai Looped",
192
+ "Noto Sans Thai Looped UI",
193
+ "Noto Sans Thai UI",
194
+ "Noto Sans Tifinagh",
195
+ "Noto Sans Tifinagh APT",
196
+ "Noto Sans Tifinagh Adrar",
197
+ "Noto Sans Tifinagh Agraw Imazighen",
198
+ "Noto Sans Tifinagh Ahaggar",
199
+ "Noto Sans Tifinagh Air",
200
+ "Noto Sans Tifinagh Azawagh",
201
+ "Noto Sans Tifinagh Ghat",
202
+ "Noto Sans Tifinagh Hawad",
203
+ "Noto Sans Tifinagh Rhissa Ixa",
204
+ "Noto Sans Tifinagh SIL",
205
+ "Noto Sans Tifinagh Tawellemmet",
206
+ "Noto Sans Tirhuta",
207
+ "Noto Sans Ugaritic",
208
+ "Noto Sans Vai",
209
+ "Noto Sans Vithkuqi",
210
+ "Noto Sans Wancho",
211
+ "Noto Sans Warang Citi",
212
+ "Noto Sans Yi",
213
+ "Noto Sans Zanabazar Square",
214
+ "Noto Traditional Nushu",
215
+ "P052",
216
+ "Sans",
217
+ "Serif",
218
+ "Source Code Pro",
219
+ "System-ui",
220
+ "URW Bookman",
221
+ "URW Gothic",
222
+ "Ubuntu",
223
+ "Ubuntu Condensed",
224
+ "Ubuntu Mono",
225
+ "Z003"
226
+ ],
227
+ "max_seq_length": 529,
228
+ "pad_size": 3,
229
+ "pixels_per_patch": 16,
230
+ "rgb": false,
231
+ "text_renderer_type": "PangoCairoTextRenderer"
232
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"unk_token": "<|endoftext|>", "bos_token": "<|endoftext|>", "eos_token": "<|endoftext|>", "add_prefix_space": true, "model_max_length": 1024, "special_tokens_map_file": null, "name_or_path": "gpt2", "tokenizer_class": "GPT2Tokenizer"}