Upload DepthAnythingForDepthEstimation
Browse files- config.json +4 -73
config.json
CHANGED
|
@@ -3,59 +3,16 @@
|
|
| 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": 1024,
|
| 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": 16,
|
| 54 |
-
"num_beam_groups": 1,
|
| 55 |
-
"num_beams": 1,
|
| 56 |
-
"num_channels": 3,
|
| 57 |
"num_hidden_layers": 24,
|
| 58 |
-
"num_return_sequences": 1,
|
| 59 |
"out_features": [
|
| 60 |
"stage21",
|
| 61 |
"stage22",
|
|
@@ -68,21 +25,8 @@
|
|
| 68 |
23,
|
| 69 |
24
|
| 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 |
-
"return_dict": true,
|
| 84 |
-
"return_dict_in_generate": false,
|
| 85 |
-
"sep_token_id": null,
|
| 86 |
"stage_names": [
|
| 87 |
"stem",
|
| 88 |
"stage1",
|
|
@@ -110,21 +54,7 @@
|
|
| 110 |
"stage23",
|
| 111 |
"stage24"
|
| 112 |
],
|
| 113 |
-
"
|
| 114 |
-
"task_specific_params": null,
|
| 115 |
-
"temperature": 1.0,
|
| 116 |
-
"tf_legacy_loss": false,
|
| 117 |
-
"tie_encoder_decoder": false,
|
| 118 |
-
"tie_word_embeddings": true,
|
| 119 |
-
"tokenizer_class": null,
|
| 120 |
-
"top_k": 50,
|
| 121 |
-
"top_p": 1.0,
|
| 122 |
-
"torch_dtype": "float32",
|
| 123 |
-
"torchscript": false,
|
| 124 |
-
"transformers_version": "4.38.0.dev0",
|
| 125 |
-
"typical_p": 1.0,
|
| 126 |
-
"use_bfloat16": false,
|
| 127 |
-
"use_swiglu_ffn": false
|
| 128 |
},
|
| 129 |
"fusion_hidden_size": 256,
|
| 130 |
"head_hidden_size": 32,
|
|
@@ -146,5 +76,6 @@
|
|
| 146 |
],
|
| 147 |
"reassemble_hidden_size": 1024,
|
| 148 |
"torch_dtype": "float32",
|
| 149 |
-
"transformers_version": null
|
|
|
|
| 150 |
}
|
|
|
|
| 3 |
"architectures": [
|
| 4 |
"DepthAnythingForDepthEstimation"
|
| 5 |
],
|
| 6 |
+
"backbone": null,
|
| 7 |
"backbone_config": {
|
|
|
|
|
|
|
|
|
|
| 8 |
"architectures": [
|
| 9 |
"Dinov2Model"
|
| 10 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
"hidden_size": 1024,
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
"image_size": 518,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
"model_type": "dinov2",
|
|
|
|
| 14 |
"num_attention_heads": 16,
|
|
|
|
|
|
|
|
|
|
| 15 |
"num_hidden_layers": 24,
|
|
|
|
| 16 |
"out_features": [
|
| 17 |
"stage21",
|
| 18 |
"stage22",
|
|
|
|
| 25 |
23,
|
| 26 |
24
|
| 27 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28 |
"patch_size": 14,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
"reshape_hidden_states": false,
|
|
|
|
|
|
|
|
|
|
| 30 |
"stage_names": [
|
| 31 |
"stem",
|
| 32 |
"stage1",
|
|
|
|
| 54 |
"stage23",
|
| 55 |
"stage24"
|
| 56 |
],
|
| 57 |
+
"torch_dtype": "float32"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 58 |
},
|
| 59 |
"fusion_hidden_size": 256,
|
| 60 |
"head_hidden_size": 32,
|
|
|
|
| 76 |
],
|
| 77 |
"reassemble_hidden_size": 1024,
|
| 78 |
"torch_dtype": "float32",
|
| 79 |
+
"transformers_version": null,
|
| 80 |
+
"use_pretrained_backbone": false
|
| 81 |
}
|