ydshieh commited on
Commit
685ba67
·
1 Parent(s): d4d5978

update after integration

Browse files
config.json CHANGED
@@ -10,7 +10,7 @@
10
  "activation_function": "gelu",
11
  "add_cross_attention": false,
12
  "architectures": null,
13
- "attention_dropout": 0,
14
  "attention_heads": 16,
15
  "bad_words_ids": null,
16
  "begin_suppress_tokens": null,
@@ -18,12 +18,10 @@
18
  "chunk_size_feed_forward": 0,
19
  "cross_attention_hidden_size": null,
20
  "decoder_start_token_id": null,
21
- "diversity_penalty": 0.0,
22
- "do_sample": false,
23
- "dropout": 0.1,
24
  "early_stopping": false,
25
  "embed_dim": 1536,
26
- "encoder_no_repeat_ngram_size": 0,
27
  "eos_token_id": 2,
28
  "exponential_decay_length_penalty": null,
29
  "ffn_dim": 6144,
@@ -44,46 +42,36 @@
44
  "layer_norm_eps": 1e-05,
45
  "layerdrop": 0.0,
46
  "layers": 24,
47
- "length_penalty": 1.0,
48
  "max_length": 20,
49
  "max_position_embeddings": 4096,
50
  "min_length": 0,
51
  "model_type": "kosmos_2_5_text_model",
52
- "no_repeat_ngram_size": 3,
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": 1,
60
  "prefix": null,
61
  "problem_type": null,
62
  "pruned_heads": {},
63
  "remove_invalid_values": false,
64
- "repetition_penalty": 1.0,
65
  "return_dict": true,
66
  "return_dict_in_generate": false,
67
  "scale_embedding": true,
68
  "sep_token_id": null,
69
  "suppress_tokens": null,
70
  "task_specific_params": null,
71
- "temperature": 1.0,
72
  "tf_legacy_loss": false,
73
  "tie_encoder_decoder": false,
74
  "tie_word_embeddings": true,
75
  "tokenizer_class": null,
76
- "top_k": 50,
77
- "top_p": 1.0,
78
  "torch_dtype": null,
79
  "torchscript": false,
80
- "typical_p": 1.0,
81
  "use_bfloat16": false,
82
  "use_cache": true,
83
  "vocab_size": 108481
84
  },
85
  "torch_dtype": "float32",
86
- "transformers_version": "4.43.3",
87
  "vision_config": {
88
  "_name_or_path": "",
89
  "add_cross_attention": false,
@@ -94,8 +82,8 @@
94
  "bos_token_id": null,
95
  "chunk_size_feed_forward": 0,
96
  "cross_attention_hidden_size": null,
97
- "d_ff": 3968,
98
- "d_kv": 64,
99
  "decoder_start_token_id": null,
100
  "dense_act_fn": "gelu_new",
101
  "diversity_penalty": 0.0,
@@ -123,7 +111,7 @@
123
  },
124
  "layer_norm_eps": 1e-06,
125
  "length_penalty": 1.0,
126
- "max_length": 20,
127
  "min_length": 0,
128
  "model_type": "kosmos_2_5_vision_model",
129
  "no_repeat_ngram_size": 0,
@@ -140,14 +128,12 @@
140
  "prefix": null,
141
  "problem_type": null,
142
  "pruned_heads": {},
143
- "relative_attention_max_distance": 128,
144
- "relative_attention_num_buckets": 32,
145
  "remove_invalid_values": false,
146
  "repetition_penalty": 1.0,
147
  "return_dict": true,
148
  "return_dict_in_generate": false,
149
  "sep_token_id": null,
150
- "seq_len": 4096,
151
  "suppress_tokens": null,
152
  "task_specific_params": null,
153
  "temperature": 1.0,
@@ -162,4 +148,4 @@
162
  "typical_p": 1.0,
163
  "use_bfloat16": false
164
  }
165
- }
 
10
  "activation_function": "gelu",
11
  "add_cross_attention": false,
12
  "architectures": null,
13
+ "attention_dropout": 0.0,
14
  "attention_heads": 16,
15
  "bad_words_ids": null,
16
  "begin_suppress_tokens": null,
 
18
  "chunk_size_feed_forward": 0,
19
  "cross_attention_hidden_size": null,
20
  "decoder_start_token_id": null,
21
+ "dropout": 0,
 
 
22
  "early_stopping": false,
23
  "embed_dim": 1536,
24
+ "pad_token_id": 1,
25
  "eos_token_id": 2,
26
  "exponential_decay_length_penalty": null,
27
  "ffn_dim": 6144,
 
42
  "layer_norm_eps": 1e-05,
43
  "layerdrop": 0.0,
44
  "layers": 24,
 
45
  "max_length": 20,
46
  "max_position_embeddings": 4096,
47
  "min_length": 0,
48
  "model_type": "kosmos_2_5_text_model",
 
 
 
49
  "num_return_sequences": 1,
50
  "output_attentions": false,
51
  "output_hidden_states": false,
52
  "output_scores": false,
 
53
  "prefix": null,
54
  "problem_type": null,
55
  "pruned_heads": {},
56
  "remove_invalid_values": false,
 
57
  "return_dict": true,
58
  "return_dict_in_generate": false,
59
  "scale_embedding": true,
60
  "sep_token_id": null,
61
  "suppress_tokens": null,
62
  "task_specific_params": null,
 
63
  "tf_legacy_loss": false,
64
  "tie_encoder_decoder": false,
65
  "tie_word_embeddings": true,
66
  "tokenizer_class": null,
 
 
67
  "torch_dtype": null,
68
  "torchscript": false,
 
69
  "use_bfloat16": false,
70
  "use_cache": true,
71
  "vocab_size": 108481
72
  },
73
  "torch_dtype": "float32",
74
+ "transformers_version": "4.42.0.dev0",
75
  "vision_config": {
76
  "_name_or_path": "",
77
  "add_cross_attention": false,
 
82
  "bos_token_id": null,
83
  "chunk_size_feed_forward": 0,
84
  "cross_attention_hidden_size": null,
85
+ "intermediate_size": 3968,
86
+ "head_dim": 64,
87
  "decoder_start_token_id": null,
88
  "dense_act_fn": "gelu_new",
89
  "diversity_penalty": 0.0,
 
111
  },
112
  "layer_norm_eps": 1e-06,
113
  "length_penalty": 1.0,
114
+ "max_length": 4096,
115
  "min_length": 0,
116
  "model_type": "kosmos_2_5_vision_model",
117
  "no_repeat_ngram_size": 0,
 
128
  "prefix": null,
129
  "problem_type": null,
130
  "pruned_heads": {},
 
 
131
  "remove_invalid_values": false,
132
  "repetition_penalty": 1.0,
133
  "return_dict": true,
134
  "return_dict_in_generate": false,
135
  "sep_token_id": null,
136
+ "max_num_patches": 4096,
137
  "suppress_tokens": null,
138
  "task_specific_params": null,
139
  "temperature": 1.0,
 
148
  "typical_p": 1.0,
149
  "use_bfloat16": false
150
  }
151
+ }
generation_config.json CHANGED
@@ -4,5 +4,5 @@
4
  "eos_token_id": 2,
5
  "no_repeat_ngram_size": 3,
6
  "pad_token_id": 1,
7
- "transformers_version": "4.43.3"
8
  }
 
4
  "eos_token_id": 2,
5
  "no_repeat_ngram_size": 3,
6
  "pad_token_id": 1,
7
+ "transformers_version": "4.43.3",
8
  }
special_tokens_map.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "boi_token": "<image>",
3
+ "bos_token": {
4
+ "content": "<s>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false
9
+ },
10
+ "eoi_token": "</image>",
11
+ "eos_token": {
12
+ "content": "</s>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false
17
+ },
18
+ "image_token": "<s>",
19
+ "pad_token": {
20
+ "content": "<pad>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false
25
+ },
26
+ "unk_token": {
27
+ "content": "<unk>",
28
+ "lstrip": false,
29
+ "normalized": false,
30
+ "rstrip": false,
31
+ "single_word": false
32
+ }
33
+ }
tokenizer.json CHANGED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json CHANGED
The diff for this file is too large to render. See raw diff