| { | |
| "architectures": [ | |
| "TraceAnythingModel" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_trace_anything.TraceAnythingConfig", | |
| "AutoModel": "modeling_trace_anything.TraceAnythingModel" | |
| }, | |
| "decoder_args": { | |
| "attn_drop": 0.0, | |
| "attn_implementation": "flash_attention", | |
| "decoder_type": "fast3r", | |
| "depth": 24, | |
| "drop": 0.0, | |
| "embed_dim": 1024, | |
| "enc_embed_dim": 1024, | |
| "mlp_ratio": 4.0, | |
| "num_heads": 16, | |
| "qkv_bias": true, | |
| "random_image_idx_embedding": false | |
| }, | |
| "dtype": "float32", | |
| "encoder_args": { | |
| "attn_implementation": "flash_attention", | |
| "depth": 24, | |
| "embed_dim": 1024, | |
| "encoder_type": "croco", | |
| "img_size": 512, | |
| "mlp_ratio": 4, | |
| "num_heads": 16, | |
| "patch_embed_cls": "ManyAR_PatchEmbed", | |
| "patch_size": 16, | |
| "pos_embed": "RoPE100" | |
| }, | |
| "head_args": { | |
| "conf_mode": [ | |
| "exp", | |
| 1, | |
| { | |
| "__float__": "Infinity" | |
| } | |
| ], | |
| "depth_mode": [ | |
| "exp", | |
| { | |
| "__float__": "-Infinity" | |
| }, | |
| { | |
| "__float__": "Infinity" | |
| } | |
| ], | |
| "head_type": "dpt", | |
| "landscape_only": true, | |
| "output_mode": "pts3d", | |
| "patch_size": 16, | |
| "with_local_head": false | |
| }, | |
| "image_size": 512, | |
| "model_type": "trace_anything", | |
| "poly_degree": 10, | |
| "targeting_mechanism": "bspline_conf", | |
| "transformers_version": "5.20.2" | |
| } | |