config.json
2.4 KB · 85 lines · json Raw
1 {
2 "_name_or_path": "florence2",
3 "architectures": [
4 "Florence2ForConditionalGeneration"
5 ],
6 "auto_map": {
7 "AutoConfig": "configuration_florence2.Florence2Config",
8 "AutoModelForCausalLM": "modeling_florence2.Florence2ForConditionalGeneration"
9 },
10 "bos_token_id": 0,
11 "eos_token_id": 2,
12 "ignore_index": -100,
13 "model_type": "florence2",
14 "pad_token_id": 1,
15 "projection_dim": 768,
16 "text_config": {
17 "vocab_size": 51289,
18 "activation_dropout": 0.1,
19 "activation_function": "gelu",
20 "add_bias_logits": false,
21 "add_final_layer_norm": false,
22 "attention_dropout": 0.1,
23 "bos_token_id": 0,
24 "classif_dropout": 0.1,
25 "classifier_dropout": 0.0,
26 "d_model": 768,
27 "decoder_attention_heads": 12,
28 "decoder_ffn_dim": 3072,
29 "decoder_layerdrop": 0.0,
30 "decoder_layers": 6,
31 "decoder_start_token_id": 2,
32 "dropout": 0.1,
33 "early_stopping": true,
34 "encoder_attention_heads": 12,
35 "encoder_ffn_dim": 3072,
36 "encoder_layerdrop": 0.0,
37 "encoder_layers": 6,
38 "eos_token_id": 2,
39 "forced_eos_token_id": 2,
40 "forced_bos_token_id": 0,
41 "gradient_checkpointing": false,
42 "init_std": 0.02,
43 "is_encoder_decoder": true,
44 "label2id": {
45 "LABEL_0": 0,
46 "LABEL_1": 1,
47 "LABEL_2": 2
48 },
49 "max_position_embeddings": 1024,
50 "no_repeat_ngram_size": 3,
51 "normalize_before": false,
52 "num_hidden_layers": 6,
53 "pad_token_id": 1,
54 "scale_embedding": false,
55 "num_beams": 3
56 },
57 "vision_config": {
58 "model_type": "davit",
59 "drop_path_rate": 0.1,
60 "patch_size": [7, 3, 3, 3],
61 "patch_stride": [4, 2, 2, 2],
62 "patch_padding": [3, 1, 1, 1],
63 "patch_prenorm": [false, true, true, true],
64 "enable_checkpoint": false,
65 "dim_embed": [128, 256, 512, 1024],
66 "num_heads": [4, 8, 16, 32],
67 "num_groups": [4, 8, 16, 32],
68 "depths": [1, 1, 9, 1],
69 "window_size": 12,
70 "projection_dim": 768,
71 "visual_temporal_embedding": {
72 "type": "COSINE",
73 "max_temporal_embeddings": 100
74 },
75 "image_pos_embed": {
76 "type": "learned_abs_2d",
77 "max_pos_embeddings": 50
78 },
79 "image_feature_source": ["spatial_avg_pool", "temporal_avg_pool"]
80 },
81 "vocab_size": 51289,
82 "torch_dtype": "float16",
83 "transformers_version": "4.41.0.dev0",
84 "is_encoder_decoder": true
85 }