Clyine1/phi3_image_question_generator

1{
2  "_name_or_path": "/content/drive/MyDrive/model/iter_39620_hf",
3  "architectures": [
4    "LlavaForConditionalGeneration"
5  ],
6  "auto_map": {
7    "AutoModelForCausalLM": "vqapair.VQApair"
8  },
9  "ignore_index": -100,
10  "image_token_index": 32011,
11  "model_type": "llava",
12  "pad_token_id": 32012,
13  "projector_hidden_act": "gelu",
14  "text_config": {
15    "_name_or_path": "../iter_32_xtuner_llama_llm",
16    "architectures": [
17      "LlamaForCausalLM"
18    ],
19    "eos_token_id": 32000,
20    "hidden_size": 3072,
21    "intermediate_size": 8192,
22    "max_position_embeddings": 4096,
23    "model_type": "llama",
24    "original_max_position_embeddings": 4096,
25    "pad_token_id": 32000,
26    "rms_norm_eps": 1e-05,
27    "sliding_window": 2047,
28    "torch_dtype": "float16",
29    "vocab_size": 32064
30  },
31  "torch_dtype": "float16",
32  "transformers_version": "4.41.2",
33  "vision_config": {
34    "_name_or_path": "./visual_encoder",
35    "architectures": [
36      "CLIPVisionModel"
37    ],
38    "dropout": 0.0,
39    "hidden_size": 1024,
40    "image_size": 336,
41    "intermediate_size": 4096,
42    "model_type": "clip_vision_model",
43    "num_attention_heads": 16,
44    "num_hidden_layers": 24,
45    "patch_size": 14,
46    "projection_dim": 768,
47    "torch_dtype": "float32"
48  },
49  "vision_feature_layer": -2,
50  "vision_feature_select_strategy": "default"
51}
52