Upload config.json
Browse files- config.json +47 -50
config.json
CHANGED
|
@@ -1,54 +1,51 @@
|
|
| 1 |
{
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
"image_token_id": 151665,
|
| 14 |
-
"initializer_range": 0.02,
|
| 15 |
-
"intermediate_size": 8960,
|
| 16 |
-
"max_position_embeddings": 131072,
|
| 17 |
-
"max_window_layers": 28,
|
| 18 |
-
"model_type": "dots_ocr",
|
| 19 |
-
"num_attention_heads": 12,
|
| 20 |
-
"num_hidden_layers": 28,
|
| 21 |
-
"num_key_value_heads": 2,
|
| 22 |
-
"rms_norm_eps": 1e-06,
|
| 23 |
-
"rope_scaling": null,
|
| 24 |
-
"rope_theta": 1000000,
|
| 25 |
-
"sliding_window": 131072,
|
| 26 |
-
"tie_word_embeddings": false,
|
| 27 |
-
"torch_dtype": "bfloat16",
|
| 28 |
-
"transformers_version": "4.51.3",
|
| 29 |
-
"use_cache": true,
|
| 30 |
-
"use_sliding_window": false,
|
| 31 |
-
"video_token_id": 151656,
|
| 32 |
-
"vision_config": {
|
| 33 |
-
"_attn_implementation_autoset": true,
|
| 34 |
-
"attn_implementation": "flash_attention_2",
|
| 35 |
-
"embed_dim": 1536,
|
| 36 |
-
"gradient_checkpointing": false,
|
| 37 |
"hidden_size": 1536,
|
| 38 |
-
"init_merger_std": 0.02,
|
| 39 |
"initializer_range": 0.02,
|
| 40 |
-
"intermediate_size":
|
| 41 |
-
"
|
| 42 |
-
"
|
| 43 |
"num_attention_heads": 12,
|
| 44 |
-
"
|
| 45 |
-
"
|
| 46 |
-
"
|
| 47 |
-
"
|
| 48 |
-
"
|
| 49 |
-
"
|
| 50 |
-
"
|
| 51 |
-
"
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"DotsOCRForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"model_type": "dots_ocr",
|
| 6 |
+
"auto_map": {
|
| 7 |
+
"AutoConfig": "configuration_dots.DotsOCRConfig",
|
| 8 |
+
"AutoModelForCausalLM": "modeling_dots_ocr.DotsOCRForCausalLM"
|
| 9 |
+
},
|
| 10 |
+
"attention_bias": true,
|
| 11 |
+
"attention_dropout": 0.0,
|
| 12 |
+
"hidden_act": "silu",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
"hidden_size": 1536,
|
|
|
|
| 14 |
"initializer_range": 0.02,
|
| 15 |
+
"intermediate_size": 8960,
|
| 16 |
+
"max_position_embeddings": 131072,
|
| 17 |
+
"max_window_layers": 28,
|
| 18 |
"num_attention_heads": 12,
|
| 19 |
+
"num_hidden_layers": 28,
|
| 20 |
+
"num_key_value_heads": 2,
|
| 21 |
+
"rms_norm_eps": 1e-06,
|
| 22 |
+
"rope_scaling": null,
|
| 23 |
+
"rope_theta": 1000000,
|
| 24 |
+
"sliding_window": 131072,
|
| 25 |
+
"tie_word_embeddings": false,
|
| 26 |
+
"torch_dtype": "bfloat16",
|
| 27 |
+
"transformers_version": "4.51.0",
|
| 28 |
+
"use_cache": true,
|
| 29 |
+
"use_sliding_window": false,
|
| 30 |
+
"vocab_size": 151936,
|
| 31 |
+
"image_token_id": 151665,
|
| 32 |
+
"video_token_id": 151656,
|
| 33 |
+
"vision_config": {
|
| 34 |
+
"embed_dim": 1536,
|
| 35 |
+
"hidden_size": 1536,
|
| 36 |
+
"intermediate_size": 4224,
|
| 37 |
+
"num_hidden_layers": 42,
|
| 38 |
+
"num_attention_heads": 12,
|
| 39 |
+
"num_channels": 3,
|
| 40 |
+
"patch_size": 14,
|
| 41 |
+
"post_norm": true,
|
| 42 |
+
"rms_norm_eps": 1e-05,
|
| 43 |
+
"spatial_merge_size": 2,
|
| 44 |
+
"temporal_patch_size": 1,
|
| 45 |
+
"use_bias": false,
|
| 46 |
+
"attn_implementation": "flash_attention_2",
|
| 47 |
+
"init_merger_std": 0.02,
|
| 48 |
+
"initializer_range": 0.02,
|
| 49 |
+
"is_causal": false
|
| 50 |
+
}
|
| 51 |
+
}
|