Upload config.json with huggingface_hub
Browse files- config.json +301 -0
config.json
ADDED
|
@@ -0,0 +1,301 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_commit_hash": null,
|
| 3 |
+
"_name_or_path": "/nas/shared/NLP_A100/wuzhenyu/ckpt/internvl_4b_20240925_stage2_grounding_dynamic_6_bs_1024_lr_4e-5_mixture",
|
| 4 |
+
"architectures": [
|
| 5 |
+
"InternVLChatModel"
|
| 6 |
+
],
|
| 7 |
+
"auto_map": {
|
| 8 |
+
"AutoConfig": "configuration_internvl_chat.InternVLChatConfig",
|
| 9 |
+
"AutoModel": "modeling_internvl_chat.InternVLChatModel",
|
| 10 |
+
"AutoModelForCausalLM": "modeling_internvl_chat.InternVLChatModel"
|
| 11 |
+
},
|
| 12 |
+
"downsample_ratio": 0.5,
|
| 13 |
+
"dynamic_image_size": true,
|
| 14 |
+
"force_image_size": 448,
|
| 15 |
+
"llm_config": {
|
| 16 |
+
"_name_or_path": "./pretrained/Phi-3-mini-128k-instruct",
|
| 17 |
+
"add_cross_attention": false,
|
| 18 |
+
"architectures": [
|
| 19 |
+
"Phi3ForCausalLM"
|
| 20 |
+
],
|
| 21 |
+
"attention_dropout": 0.0,
|
| 22 |
+
"auto_map": {
|
| 23 |
+
"AutoConfig": "configuration_phi3.Phi3Config",
|
| 24 |
+
"AutoModelForCausalLM": "modeling_phi3.Phi3ForCausalLM"
|
| 25 |
+
},
|
| 26 |
+
"bad_words_ids": null,
|
| 27 |
+
"begin_suppress_tokens": null,
|
| 28 |
+
"bos_token_id": 1,
|
| 29 |
+
"chunk_size_feed_forward": 0,
|
| 30 |
+
"cross_attention_hidden_size": null,
|
| 31 |
+
"decoder_start_token_id": null,
|
| 32 |
+
"diversity_penalty": 0.0,
|
| 33 |
+
"do_sample": false,
|
| 34 |
+
"early_stopping": false,
|
| 35 |
+
"embd_pdrop": 0.0,
|
| 36 |
+
"encoder_no_repeat_ngram_size": 0,
|
| 37 |
+
"eos_token_id": 32000,
|
| 38 |
+
"exponential_decay_length_penalty": null,
|
| 39 |
+
"finetuning_task": null,
|
| 40 |
+
"forced_bos_token_id": null,
|
| 41 |
+
"forced_eos_token_id": null,
|
| 42 |
+
"hidden_act": "silu",
|
| 43 |
+
"hidden_size": 3072,
|
| 44 |
+
"id2label": {
|
| 45 |
+
"0": "LABEL_0",
|
| 46 |
+
"1": "LABEL_1"
|
| 47 |
+
},
|
| 48 |
+
"initializer_range": 0.02,
|
| 49 |
+
"intermediate_size": 8192,
|
| 50 |
+
"is_decoder": false,
|
| 51 |
+
"is_encoder_decoder": false,
|
| 52 |
+
"label2id": {
|
| 53 |
+
"LABEL_0": 0,
|
| 54 |
+
"LABEL_1": 1
|
| 55 |
+
},
|
| 56 |
+
"length_penalty": 1.0,
|
| 57 |
+
"max_length": 20,
|
| 58 |
+
"max_position_embeddings": 131072,
|
| 59 |
+
"min_length": 0,
|
| 60 |
+
"model_type": "phi3",
|
| 61 |
+
"no_repeat_ngram_size": 0,
|
| 62 |
+
"num_attention_heads": 32,
|
| 63 |
+
"num_beam_groups": 1,
|
| 64 |
+
"num_beams": 1,
|
| 65 |
+
"num_hidden_layers": 32,
|
| 66 |
+
"num_key_value_heads": 32,
|
| 67 |
+
"num_return_sequences": 1,
|
| 68 |
+
"original_max_position_embeddings": 4096,
|
| 69 |
+
"output_attentions": false,
|
| 70 |
+
"output_hidden_states": false,
|
| 71 |
+
"output_scores": false,
|
| 72 |
+
"pad_token_id": 32000,
|
| 73 |
+
"prefix": null,
|
| 74 |
+
"problem_type": null,
|
| 75 |
+
"pruned_heads": {},
|
| 76 |
+
"remove_invalid_values": false,
|
| 77 |
+
"repetition_penalty": 1.0,
|
| 78 |
+
"resid_pdrop": 0.0,
|
| 79 |
+
"return_dict": true,
|
| 80 |
+
"return_dict_in_generate": false,
|
| 81 |
+
"rms_norm_eps": 1e-05,
|
| 82 |
+
"rope_scaling": {
|
| 83 |
+
"long_factor": [
|
| 84 |
+
1.0299999713897705,
|
| 85 |
+
1.0499999523162842,
|
| 86 |
+
1.0499999523162842,
|
| 87 |
+
1.0799999237060547,
|
| 88 |
+
1.2299998998641968,
|
| 89 |
+
1.2299998998641968,
|
| 90 |
+
1.2999999523162842,
|
| 91 |
+
1.4499999284744263,
|
| 92 |
+
1.5999999046325684,
|
| 93 |
+
1.6499998569488525,
|
| 94 |
+
1.8999998569488525,
|
| 95 |
+
2.859999895095825,
|
| 96 |
+
3.68999981880188,
|
| 97 |
+
5.419999599456787,
|
| 98 |
+
5.489999771118164,
|
| 99 |
+
5.489999771118164,
|
| 100 |
+
9.09000015258789,
|
| 101 |
+
11.579999923706055,
|
| 102 |
+
15.65999984741211,
|
| 103 |
+
15.769999504089355,
|
| 104 |
+
15.789999961853027,
|
| 105 |
+
18.360000610351562,
|
| 106 |
+
21.989999771118164,
|
| 107 |
+
23.079999923706055,
|
| 108 |
+
30.009998321533203,
|
| 109 |
+
32.35000228881836,
|
| 110 |
+
32.590003967285156,
|
| 111 |
+
35.56000518798828,
|
| 112 |
+
39.95000457763672,
|
| 113 |
+
53.840003967285156,
|
| 114 |
+
56.20000457763672,
|
| 115 |
+
57.95000457763672,
|
| 116 |
+
59.29000473022461,
|
| 117 |
+
59.77000427246094,
|
| 118 |
+
59.920005798339844,
|
| 119 |
+
61.190006256103516,
|
| 120 |
+
61.96000671386719,
|
| 121 |
+
62.50000762939453,
|
| 122 |
+
63.3700065612793,
|
| 123 |
+
63.48000717163086,
|
| 124 |
+
63.48000717163086,
|
| 125 |
+
63.66000747680664,
|
| 126 |
+
63.850006103515625,
|
| 127 |
+
64.08000946044922,
|
| 128 |
+
64.760009765625,
|
| 129 |
+
64.80001068115234,
|
| 130 |
+
64.81001281738281,
|
| 131 |
+
64.81001281738281
|
| 132 |
+
],
|
| 133 |
+
"short_factor": [
|
| 134 |
+
1.05,
|
| 135 |
+
1.05,
|
| 136 |
+
1.05,
|
| 137 |
+
1.1,
|
| 138 |
+
1.1,
|
| 139 |
+
1.1500000000000001,
|
| 140 |
+
1.2000000000000002,
|
| 141 |
+
1.2500000000000002,
|
| 142 |
+
1.3000000000000003,
|
| 143 |
+
1.3500000000000003,
|
| 144 |
+
1.5000000000000004,
|
| 145 |
+
2.000000000000001,
|
| 146 |
+
2.000000000000001,
|
| 147 |
+
2.000000000000001,
|
| 148 |
+
2.000000000000001,
|
| 149 |
+
2.000000000000001,
|
| 150 |
+
2.000000000000001,
|
| 151 |
+
2.000000000000001,
|
| 152 |
+
2.000000000000001,
|
| 153 |
+
2.000000000000001,
|
| 154 |
+
2.000000000000001,
|
| 155 |
+
2.000000000000001,
|
| 156 |
+
2.000000000000001,
|
| 157 |
+
2.000000000000001,
|
| 158 |
+
2.000000000000001,
|
| 159 |
+
2.000000000000001,
|
| 160 |
+
2.000000000000001,
|
| 161 |
+
2.000000000000001,
|
| 162 |
+
2.000000000000001,
|
| 163 |
+
2.000000000000001,
|
| 164 |
+
2.000000000000001,
|
| 165 |
+
2.000000000000001,
|
| 166 |
+
2.0500000000000007,
|
| 167 |
+
2.0500000000000007,
|
| 168 |
+
2.0500000000000007,
|
| 169 |
+
2.1000000000000005,
|
| 170 |
+
2.1000000000000005,
|
| 171 |
+
2.1000000000000005,
|
| 172 |
+
2.1500000000000004,
|
| 173 |
+
2.1500000000000004,
|
| 174 |
+
2.3499999999999996,
|
| 175 |
+
2.549999999999999,
|
| 176 |
+
2.5999999999999988,
|
| 177 |
+
2.5999999999999988,
|
| 178 |
+
2.7499999999999982,
|
| 179 |
+
2.849999999999998,
|
| 180 |
+
2.849999999999998,
|
| 181 |
+
2.9499999999999975
|
| 182 |
+
],
|
| 183 |
+
"type": "su"
|
| 184 |
+
},
|
| 185 |
+
"rope_theta": 10000.0,
|
| 186 |
+
"sep_token_id": null,
|
| 187 |
+
"sliding_window": 262144,
|
| 188 |
+
"suppress_tokens": null,
|
| 189 |
+
"task_specific_params": null,
|
| 190 |
+
"temperature": 1.0,
|
| 191 |
+
"tf_legacy_loss": false,
|
| 192 |
+
"tie_encoder_decoder": false,
|
| 193 |
+
"tie_word_embeddings": false,
|
| 194 |
+
"tokenizer_class": null,
|
| 195 |
+
"top_k": 50,
|
| 196 |
+
"top_p": null,
|
| 197 |
+
"torch_dtype": "bfloat16",
|
| 198 |
+
"torchscript": false,
|
| 199 |
+
"transformers_version": "4.41.2",
|
| 200 |
+
"typical_p": 1.0,
|
| 201 |
+
"use_bfloat16": true,
|
| 202 |
+
"use_cache": false,
|
| 203 |
+
"vocab_size": 32022
|
| 204 |
+
},
|
| 205 |
+
"max_dynamic_patch": 6,
|
| 206 |
+
"min_dynamic_patch": 1,
|
| 207 |
+
"model_type": "internvl_chat",
|
| 208 |
+
"pad2square": false,
|
| 209 |
+
"ps_version": "v2",
|
| 210 |
+
"select_layer": -1,
|
| 211 |
+
"template": "phi3-chat",
|
| 212 |
+
"torch_dtype": "bfloat16",
|
| 213 |
+
"transformers_version": null,
|
| 214 |
+
"use_backbone_lora": 0,
|
| 215 |
+
"use_llm_lora": 0,
|
| 216 |
+
"use_thumbnail": true,
|
| 217 |
+
"vision_config": {
|
| 218 |
+
"_name_or_path": "",
|
| 219 |
+
"add_cross_attention": false,
|
| 220 |
+
"architectures": [
|
| 221 |
+
"InternVisionModel"
|
| 222 |
+
],
|
| 223 |
+
"attention_dropout": 0.0,
|
| 224 |
+
"bad_words_ids": null,
|
| 225 |
+
"begin_suppress_tokens": null,
|
| 226 |
+
"bos_token_id": null,
|
| 227 |
+
"chunk_size_feed_forward": 0,
|
| 228 |
+
"cross_attention_hidden_size": null,
|
| 229 |
+
"decoder_start_token_id": null,
|
| 230 |
+
"diversity_penalty": 0.0,
|
| 231 |
+
"do_sample": false,
|
| 232 |
+
"drop_path_rate": 0.1,
|
| 233 |
+
"dropout": 0.0,
|
| 234 |
+
"early_stopping": false,
|
| 235 |
+
"encoder_no_repeat_ngram_size": 0,
|
| 236 |
+
"eos_token_id": null,
|
| 237 |
+
"exponential_decay_length_penalty": null,
|
| 238 |
+
"finetuning_task": null,
|
| 239 |
+
"forced_bos_token_id": null,
|
| 240 |
+
"forced_eos_token_id": null,
|
| 241 |
+
"hidden_act": "gelu",
|
| 242 |
+
"hidden_size": 1024,
|
| 243 |
+
"id2label": {
|
| 244 |
+
"0": "LABEL_0",
|
| 245 |
+
"1": "LABEL_1"
|
| 246 |
+
},
|
| 247 |
+
"image_size": 448,
|
| 248 |
+
"initializer_factor": 1.0,
|
| 249 |
+
"initializer_range": 0.02,
|
| 250 |
+
"intermediate_size": 4096,
|
| 251 |
+
"is_decoder": false,
|
| 252 |
+
"is_encoder_decoder": false,
|
| 253 |
+
"label2id": {
|
| 254 |
+
"LABEL_0": 0,
|
| 255 |
+
"LABEL_1": 1
|
| 256 |
+
},
|
| 257 |
+
"layer_norm_eps": 1e-06,
|
| 258 |
+
"length_penalty": 1.0,
|
| 259 |
+
"max_length": 20,
|
| 260 |
+
"min_length": 0,
|
| 261 |
+
"model_type": "intern_vit_6b",
|
| 262 |
+
"no_repeat_ngram_size": 0,
|
| 263 |
+
"norm_type": "layer_norm",
|
| 264 |
+
"num_attention_heads": 16,
|
| 265 |
+
"num_beam_groups": 1,
|
| 266 |
+
"num_beams": 1,
|
| 267 |
+
"num_channels": 3,
|
| 268 |
+
"num_hidden_layers": 24,
|
| 269 |
+
"num_return_sequences": 1,
|
| 270 |
+
"output_attentions": false,
|
| 271 |
+
"output_hidden_states": false,
|
| 272 |
+
"output_scores": false,
|
| 273 |
+
"pad_token_id": null,
|
| 274 |
+
"patch_size": 14,
|
| 275 |
+
"prefix": null,
|
| 276 |
+
"problem_type": null,
|
| 277 |
+
"pruned_heads": {},
|
| 278 |
+
"qk_normalization": false,
|
| 279 |
+
"qkv_bias": true,
|
| 280 |
+
"remove_invalid_values": false,
|
| 281 |
+
"repetition_penalty": 1.0,
|
| 282 |
+
"return_dict": true,
|
| 283 |
+
"return_dict_in_generate": false,
|
| 284 |
+
"sep_token_id": null,
|
| 285 |
+
"suppress_tokens": null,
|
| 286 |
+
"task_specific_params": null,
|
| 287 |
+
"temperature": 1.0,
|
| 288 |
+
"tf_legacy_loss": false,
|
| 289 |
+
"tie_encoder_decoder": false,
|
| 290 |
+
"tie_word_embeddings": true,
|
| 291 |
+
"tokenizer_class": null,
|
| 292 |
+
"top_k": 50,
|
| 293 |
+
"top_p": null,
|
| 294 |
+
"torch_dtype": "bfloat16",
|
| 295 |
+
"torchscript": false,
|
| 296 |
+
"transformers_version": "4.41.2",
|
| 297 |
+
"typical_p": 1.0,
|
| 298 |
+
"use_bfloat16": true,
|
| 299 |
+
"use_flash_attn": true
|
| 300 |
+
}
|
| 301 |
+
}
|