Add model config.json
Browse files- config.json +96 -0
config.json
ADDED
|
@@ -0,0 +1,96 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"Qwen2_5_VLForConditionalGeneration"
|
| 4 |
+
],
|
| 5 |
+
"attention_dropout": 0.0,
|
| 6 |
+
"bos_token_id": 151643,
|
| 7 |
+
"eos_token_id": 151645,
|
| 8 |
+
"vision_start_token_id": 151652,
|
| 9 |
+
"vision_end_token_id": 151653,
|
| 10 |
+
"vision_token_id": 151654,
|
| 11 |
+
"image_token_id": 151655,
|
| 12 |
+
"video_token_id": 151656,
|
| 13 |
+
"hidden_act": "silu",
|
| 14 |
+
"hidden_size": 2048,
|
| 15 |
+
"action_hidden_size": 1024,
|
| 16 |
+
"state_hidden_size": 2048,
|
| 17 |
+
"initializer_range": 0.02,
|
| 18 |
+
"intermediate_size": 11008,
|
| 19 |
+
"max_position_embeddings": 128000,
|
| 20 |
+
"max_window_layers": 70,
|
| 21 |
+
"model_type": "qwen2_5_vl",
|
| 22 |
+
"num_attention_heads": 16,
|
| 23 |
+
"num_hidden_layers": 36,
|
| 24 |
+
"num_key_value_heads": 2,
|
| 25 |
+
"rms_norm_eps": 1e-06,
|
| 26 |
+
"rope_theta": 1000000.0,
|
| 27 |
+
"sliding_window": 32768,
|
| 28 |
+
"tie_word_embeddings": true,
|
| 29 |
+
"torch_dtype": "bfloat16",
|
| 30 |
+
"transformers_version": "4.41.2",
|
| 31 |
+
"_attn_implementation": "flash_mask",
|
| 32 |
+
"use_cache": true,
|
| 33 |
+
"use_sliding_window": false,
|
| 34 |
+
"vision_config": {
|
| 35 |
+
"depth": 32,
|
| 36 |
+
"hidden_act": "silu",
|
| 37 |
+
"hidden_size": 1280,
|
| 38 |
+
"intermediate_size": 3420,
|
| 39 |
+
"num_heads": 16,
|
| 40 |
+
"in_chans": 3,
|
| 41 |
+
"out_hidden_size": 2048,
|
| 42 |
+
"patch_size": 14,
|
| 43 |
+
"spatial_merge_size": 2,
|
| 44 |
+
"spatial_patch_size": 14,
|
| 45 |
+
"window_size": 112,
|
| 46 |
+
"fullatt_block_indexes": [
|
| 47 |
+
7,
|
| 48 |
+
15,
|
| 49 |
+
23,
|
| 50 |
+
31
|
| 51 |
+
],
|
| 52 |
+
"tokens_per_second": 2,
|
| 53 |
+
"temporal_patch_size": 2,
|
| 54 |
+
"_attn_implementation": "flash_attention_2"
|
| 55 |
+
},
|
| 56 |
+
"rope_scaling": {
|
| 57 |
+
"type": "mrope",
|
| 58 |
+
"mrope_section": [
|
| 59 |
+
16,
|
| 60 |
+
24,
|
| 61 |
+
24
|
| 62 |
+
]
|
| 63 |
+
},
|
| 64 |
+
"vocab_size": 151936,
|
| 65 |
+
"num_experts": 2,
|
| 66 |
+
"experts":[
|
| 67 |
+
{
|
| 68 |
+
"hidden_size": 2048,
|
| 69 |
+
"intermediate_size": 11008,
|
| 70 |
+
"hidden_act": "silu"
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"hidden_size": 1024,
|
| 74 |
+
"intermediate_size": 2048,
|
| 75 |
+
"hidden_act": "silu"
|
| 76 |
+
}
|
| 77 |
+
],
|
| 78 |
+
"noise_scheduler": {
|
| 79 |
+
"beta_alpha": 1.5,
|
| 80 |
+
"beta_beta": 1.0,
|
| 81 |
+
"s": 0.999,
|
| 82 |
+
"num_inference_timesteps": 10
|
| 83 |
+
},
|
| 84 |
+
"dim_inputs": [2048,1024],
|
| 85 |
+
"attention_moe": true,
|
| 86 |
+
"attention_moe_config":[
|
| 87 |
+
|
| 88 |
+
],
|
| 89 |
+
"mlp_moe": true,
|
| 90 |
+
"norm_moe": true,
|
| 91 |
+
"mot_opt": true,
|
| 92 |
+
"flow_loss_weight": 1.0,
|
| 93 |
+
"causal_action_attention_mask": true,
|
| 94 |
+
"use_x_pred": false,
|
| 95 |
+
"use_x_loss": true
|
| 96 |
+
}
|