Upload DepthAnythingForDepthEstimation
Browse files- config.json +4 -89
config.json
CHANGED
@@ -3,59 +3,15 @@
|
|
3 |
"architectures": [
|
4 |
"DepthAnythingForDepthEstimation"
|
5 |
],
|
|
|
6 |
"backbone_config": {
|
7 |
-
"_name_or_path": "",
|
8 |
-
"add_cross_attention": false,
|
9 |
-
"apply_layernorm": true,
|
10 |
"architectures": [
|
11 |
"Dinov2Model"
|
12 |
],
|
13 |
-
"attention_probs_dropout_prob": 0.0,
|
14 |
-
"bad_words_ids": null,
|
15 |
-
"begin_suppress_tokens": null,
|
16 |
-
"bos_token_id": null,
|
17 |
-
"chunk_size_feed_forward": 0,
|
18 |
-
"cross_attention_hidden_size": null,
|
19 |
-
"decoder_start_token_id": null,
|
20 |
-
"diversity_penalty": 0.0,
|
21 |
-
"do_sample": false,
|
22 |
-
"drop_path_rate": 0.0,
|
23 |
-
"early_stopping": false,
|
24 |
-
"encoder_no_repeat_ngram_size": 0,
|
25 |
-
"eos_token_id": null,
|
26 |
-
"exponential_decay_length_penalty": null,
|
27 |
-
"finetuning_task": null,
|
28 |
-
"forced_bos_token_id": null,
|
29 |
-
"forced_eos_token_id": null,
|
30 |
-
"hidden_act": "gelu",
|
31 |
-
"hidden_dropout_prob": 0.0,
|
32 |
"hidden_size": 768,
|
33 |
-
"id2label": {
|
34 |
-
"0": "LABEL_0",
|
35 |
-
"1": "LABEL_1"
|
36 |
-
},
|
37 |
"image_size": 518,
|
38 |
-
"initializer_range": 0.02,
|
39 |
-
"is_decoder": false,
|
40 |
-
"is_encoder_decoder": false,
|
41 |
-
"label2id": {
|
42 |
-
"LABEL_0": 0,
|
43 |
-
"LABEL_1": 1
|
44 |
-
},
|
45 |
-
"layer_norm_eps": 1e-06,
|
46 |
-
"layerscale_value": 1.0,
|
47 |
-
"length_penalty": 1.0,
|
48 |
-
"max_length": 20,
|
49 |
-
"min_length": 0,
|
50 |
-
"mlp_ratio": 4,
|
51 |
"model_type": "dinov2",
|
52 |
-
"no_repeat_ngram_size": 0,
|
53 |
"num_attention_heads": 12,
|
54 |
-
"num_beam_groups": 1,
|
55 |
-
"num_beams": 1,
|
56 |
-
"num_channels": 3,
|
57 |
-
"num_hidden_layers": 12,
|
58 |
-
"num_return_sequences": 1,
|
59 |
"out_features": [
|
60 |
"stage9",
|
61 |
"stage10",
|
@@ -68,51 +24,9 @@
|
|
68 |
11,
|
69 |
12
|
70 |
],
|
71 |
-
"output_attentions": false,
|
72 |
-
"output_hidden_states": false,
|
73 |
-
"output_scores": false,
|
74 |
-
"pad_token_id": null,
|
75 |
"patch_size": 14,
|
76 |
-
"prefix": null,
|
77 |
-
"problem_type": null,
|
78 |
-
"pruned_heads": {},
|
79 |
-
"qkv_bias": true,
|
80 |
-
"remove_invalid_values": false,
|
81 |
-
"repetition_penalty": 1.0,
|
82 |
"reshape_hidden_states": false,
|
83 |
-
"
|
84 |
-
"return_dict_in_generate": false,
|
85 |
-
"sep_token_id": null,
|
86 |
-
"stage_names": [
|
87 |
-
"stem",
|
88 |
-
"stage1",
|
89 |
-
"stage2",
|
90 |
-
"stage3",
|
91 |
-
"stage4",
|
92 |
-
"stage5",
|
93 |
-
"stage6",
|
94 |
-
"stage7",
|
95 |
-
"stage8",
|
96 |
-
"stage9",
|
97 |
-
"stage10",
|
98 |
-
"stage11",
|
99 |
-
"stage12"
|
100 |
-
],
|
101 |
-
"suppress_tokens": null,
|
102 |
-
"task_specific_params": null,
|
103 |
-
"temperature": 1.0,
|
104 |
-
"tf_legacy_loss": false,
|
105 |
-
"tie_encoder_decoder": false,
|
106 |
-
"tie_word_embeddings": true,
|
107 |
-
"tokenizer_class": null,
|
108 |
-
"top_k": 50,
|
109 |
-
"top_p": 1.0,
|
110 |
-
"torch_dtype": "float32",
|
111 |
-
"torchscript": false,
|
112 |
-
"transformers_version": "4.38.0.dev0",
|
113 |
-
"typical_p": 1.0,
|
114 |
-
"use_bfloat16": false,
|
115 |
-
"use_swiglu_ffn": false
|
116 |
},
|
117 |
"fusion_hidden_size": 128,
|
118 |
"head_hidden_size": 32,
|
@@ -134,5 +48,6 @@
|
|
134 |
],
|
135 |
"reassemble_hidden_size": 768,
|
136 |
"torch_dtype": "float32",
|
137 |
-
"transformers_version": null
|
|
|
138 |
}
|
|
|
3 |
"architectures": [
|
4 |
"DepthAnythingForDepthEstimation"
|
5 |
],
|
6 |
+
"backbone": null,
|
7 |
"backbone_config": {
|
|
|
|
|
|
|
8 |
"architectures": [
|
9 |
"Dinov2Model"
|
10 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
"hidden_size": 768,
|
|
|
|
|
|
|
|
|
12 |
"image_size": 518,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
13 |
"model_type": "dinov2",
|
|
|
14 |
"num_attention_heads": 12,
|
|
|
|
|
|
|
|
|
|
|
15 |
"out_features": [
|
16 |
"stage9",
|
17 |
"stage10",
|
|
|
24 |
11,
|
25 |
12
|
26 |
],
|
|
|
|
|
|
|
|
|
27 |
"patch_size": 14,
|
|
|
|
|
|
|
|
|
|
|
|
|
28 |
"reshape_hidden_states": false,
|
29 |
+
"torch_dtype": "float32"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
30 |
},
|
31 |
"fusion_hidden_size": 128,
|
32 |
"head_hidden_size": 32,
|
|
|
48 |
],
|
49 |
"reassemble_hidden_size": 768,
|
50 |
"torch_dtype": "float32",
|
51 |
+
"transformers_version": null,
|
52 |
+
"use_pretrained_backbone": false
|
53 |
}
|