Upload config.json with huggingface_hub
Browse files- config.json +76 -0
config.json
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_attn_implementation": "flash_attention_2",
|
| 3 |
+
"_commit_hash": null,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"Eagle25VLForConditionalGeneration"
|
| 6 |
+
],
|
| 7 |
+
"auto_map": {
|
| 8 |
+
"AutoConfig": "configuration_eagle2_5_vl.Eagle25VLConfig",
|
| 9 |
+
"AutoModel": "modeling_eagle2_5_vl.Eagle25VLForConditionalGeneration"
|
| 10 |
+
},
|
| 11 |
+
"downsample_ratio": 0.5,
|
| 12 |
+
"dynamic_image_size": true,
|
| 13 |
+
"force_image_size": 224,
|
| 14 |
+
"image_token_index": 151669,
|
| 15 |
+
"loss_version": "efficient_v2_cp_head",
|
| 16 |
+
"max_dynamic_tiles": 12,
|
| 17 |
+
"min_dynamic_tiles": 1,
|
| 18 |
+
"mlp_checkpoint": false,
|
| 19 |
+
"mlp_connector_layers": 1,
|
| 20 |
+
"model_type": "eagle_2_5_vl",
|
| 21 |
+
"pad2square": false,
|
| 22 |
+
"select_layer": -1,
|
| 23 |
+
"template": "qwen3-chat",
|
| 24 |
+
"text_config": {
|
| 25 |
+
"_attn_implementation_autoset": true,
|
| 26 |
+
"_name_or_path": "Qwen/Qwen3-1.7B",
|
| 27 |
+
"architectures": [
|
| 28 |
+
"Qwen3ForCausalLM"
|
| 29 |
+
],
|
| 30 |
+
"attention_bias": false,
|
| 31 |
+
"attention_dropout": 0,
|
| 32 |
+
"bos_token_id": 151643,
|
| 33 |
+
"eos_token_id": 151645,
|
| 34 |
+
"head_dim": 128,
|
| 35 |
+
"hidden_act": "silu",
|
| 36 |
+
"hidden_size": 2048,
|
| 37 |
+
"initializer_range": 0.02,
|
| 38 |
+
"intermediate_size": 6144,
|
| 39 |
+
"max_position_embeddings": 40960,
|
| 40 |
+
"max_window_layers": 28,
|
| 41 |
+
"model_type": "qwen3",
|
| 42 |
+
"num_attention_heads": 16,
|
| 43 |
+
"num_hidden_layers": 28,
|
| 44 |
+
"num_key_value_heads": 8,
|
| 45 |
+
"rms_norm_eps": 0.000001,
|
| 46 |
+
"rope_scaling": null,
|
| 47 |
+
"rope_theta": 1000000,
|
| 48 |
+
"sliding_window": null,
|
| 49 |
+
"tie_word_embeddings": true,
|
| 50 |
+
"torch_dtype": "bfloat16",
|
| 51 |
+
"use_cache": false,
|
| 52 |
+
"use_sliding_window": false,
|
| 53 |
+
"vocab_size": 151680
|
| 54 |
+
},
|
| 55 |
+
"torch_dtype": "bfloat16",
|
| 56 |
+
"transformers_version": null,
|
| 57 |
+
"use_backbone_lora": 0,
|
| 58 |
+
"use_llm_lora": 0,
|
| 59 |
+
"use_pixel_shuffle": false,
|
| 60 |
+
"use_thumbnail": true,
|
| 61 |
+
"vision_config": {
|
| 62 |
+
"_attn_implementation_autoset": true,
|
| 63 |
+
"attention_dropout": 0,
|
| 64 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 65 |
+
"hidden_size": 1152,
|
| 66 |
+
"image_size": 224,
|
| 67 |
+
"intermediate_size": 4304,
|
| 68 |
+
"layer_norm_eps": 0.000001,
|
| 69 |
+
"model_type": "siglip_vision_model",
|
| 70 |
+
"num_attention_heads": 16,
|
| 71 |
+
"num_channels": 3,
|
| 72 |
+
"num_hidden_layers": 27,
|
| 73 |
+
"patch_size": 14,
|
| 74 |
+
"torch_dtype": "bfloat16"
|
| 75 |
+
}
|
| 76 |
+
}
|