Upload config.json with huggingface_hub
Browse files- config.json +60 -0
config.json
ADDED
|
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "/mnt/bn/hpp-develop/repos/EcoMLLM/checkpoints/tinyvalley_b2_v3-s2v1/",
|
| 3 |
+
"anyres": true,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"ValleyQwen2ForCausalLM"
|
| 6 |
+
],
|
| 7 |
+
"attention_dropout": 0.0,
|
| 8 |
+
"bos_token_id": 151643,
|
| 9 |
+
"eagle_vision_tower": null,
|
| 10 |
+
"eos_token_id": 151645,
|
| 11 |
+
"hidden_act": "silu",
|
| 12 |
+
"hidden_size": 1536,
|
| 13 |
+
"image_aspect_ratio": null,
|
| 14 |
+
"image_grid_pinpoints": null,
|
| 15 |
+
"initializer_range": 0.02,
|
| 16 |
+
"intermediate_size": 8960,
|
| 17 |
+
"language": "english",
|
| 18 |
+
"llm_name": "qwen2",
|
| 19 |
+
"max_position_embeddings": 32768,
|
| 20 |
+
"max_tile_num": 9,
|
| 21 |
+
"max_vision_token": 320,
|
| 22 |
+
"max_window_layers": 28,
|
| 23 |
+
"min_tile_num": 1,
|
| 24 |
+
"mlp_hidden_dim": null,
|
| 25 |
+
"mm_hidden_size": 1152,
|
| 26 |
+
"mm_projector_type": "mlp_pixel_shuffle",
|
| 27 |
+
"mm_use_im_patch_token": false,
|
| 28 |
+
"mm_use_im_start_end": false,
|
| 29 |
+
"mm_vision_select_feature": "cls_patch",
|
| 30 |
+
"mm_vision_select_layer": -2,
|
| 31 |
+
"mm_vision_tower": "/mnt/bn/hpp-develop/repos/EcoMLLM/google/siglip-so400m-patch14-384/",
|
| 32 |
+
"model_class": "valley-product",
|
| 33 |
+
"model_type": "valley",
|
| 34 |
+
"navit_merger_hidden_dim": null,
|
| 35 |
+
"navit_use_mm_projector": false,
|
| 36 |
+
"num_attention_heads": 12,
|
| 37 |
+
"num_hidden_layers": 28,
|
| 38 |
+
"num_key_value_heads": 2,
|
| 39 |
+
"only_crop_single_image": false,
|
| 40 |
+
"only_navit": false,
|
| 41 |
+
"pixelshuffle_downsample_ratio": 3,
|
| 42 |
+
"pool_out_size": 8,
|
| 43 |
+
"pooling_stride": 1,
|
| 44 |
+
"rms_norm_eps": 1e-06,
|
| 45 |
+
"rope_scaling": null,
|
| 46 |
+
"rope_theta": 1000000.0,
|
| 47 |
+
"sliding_window": null,
|
| 48 |
+
"tie_word_embeddings": true,
|
| 49 |
+
"tokenize_function": null,
|
| 50 |
+
"torch_dtype": "bfloat16",
|
| 51 |
+
"training_stage": "stage2",
|
| 52 |
+
"transformers_version": "4.45.2",
|
| 53 |
+
"tune_mm_mlp_adapter": true,
|
| 54 |
+
"use_cache": true,
|
| 55 |
+
"use_sliding_window": false,
|
| 56 |
+
"use_special_start_end_token": true,
|
| 57 |
+
"vit_crop_size": 384,
|
| 58 |
+
"vit_patch_size": 14,
|
| 59 |
+
"vocab_size": 151652
|
| 60 |
+
}
|