Download gui_jepa_config.json from LilCC/GUI-JEPA: direct link, hf CLI and curl.
- Browser
- Download file 2.96 kB
-
https://huggingface.co/LilCC/GUI-JEPA/resolve/main/gui_jepa_config.json
- Command line
-
hf download hf://LilCC/GUI-JEPA/gui_jepa_config.json
-
curl -L -o gui_jepa_config.json https://huggingface.co/LilCC/GUI-JEPA/resolve/main/gui_jepa_config.json
2.96 kB
| { | |
| "format_version": 1, | |
| "model_name": "GUI-JEPA", | |
| "paper": "https://arxiv.org/abs/2610.36877", | |
| "code": "https://github.com/bcwang-sjtu/GUI-JEPA", | |
| "checkpoint_variant": "final policy, seed 3", | |
| "training_seed": 3, | |
| "evaluation_seed": 1, | |
| "encoder": { | |
| "path": "encoder/encoder_1.pt", | |
| "architecture": "vit_large_patch16_224", | |
| "initialization": "timm/vit_large_patch16_224.augreg_in21k_ft_in1k", | |
| "image_size": [ | |
| 224, | |
| 224 | |
| ], | |
| "patch_size": 16, | |
| "grid_size": [ | |
| 14, | |
| 14 | |
| ], | |
| "embedding_dim": 1024, | |
| "drop_cls_token": true, | |
| "frozen": true, | |
| "preprocessing": { | |
| "color_mode": "RGB", | |
| "resize": "PIL.Image.Image.resize((224, 224)) with the default RGB resampling; no center crop", | |
| "scale": 0.00392156862745098, | |
| "mean": [ | |
| 0.485, | |
| 0.456, | |
| 0.406 | |
| ], | |
| "std": [ | |
| 0.229, | |
| 0.224, | |
| 0.225 | |
| ] | |
| } | |
| }, | |
| "policy": { | |
| "path": "policy", | |
| "heads_file": "heads.pt", | |
| "heads_keys": [ | |
| "wm_head", | |
| "matcher_head", | |
| "predictor" | |
| ], | |
| "matcher_path": "policy/matcher_lm", | |
| "matcher_base_model": "sentence-transformers/all-MiniLM-L12-v2", | |
| "matcher_max_length": 96, | |
| "predictor_depth": 3, | |
| "predictor_attention_heads": 16, | |
| "candidate_action_dim": 16, | |
| "predictor_action_input_dim": 1040, | |
| "residual": true, | |
| "history": true, | |
| "history_text_steps": 4, | |
| "condition_extra_dim": 1040, | |
| "structural_projection_dim": 256, | |
| "action_kinds": [ | |
| "click", | |
| "input_text", | |
| "scroll", | |
| "navigate_back", | |
| "navigate_home", | |
| "wait", | |
| "open_app" | |
| ] | |
| }, | |
| "span_ranker": { | |
| "path": "span_ranker", | |
| "head_file": "ranker_head.pt", | |
| "language_model_path": "span_ranker/ranker_lm", | |
| "base_model": "sentence-transformers/all-MiniLM-L6-v2", | |
| "max_words": 16, | |
| "span_cap": 800, | |
| "source": "goal", | |
| "max_length": 96 | |
| }, | |
| "click_reranker": { | |
| "path": "click_reranker", | |
| "head_file": "reranker_head.pt", | |
| "language_model_path": "click_reranker/reranker_lm", | |
| "base_model": "sentence-transformers/all-MiniLM-L12-v2", | |
| "aux_dim": 11, | |
| "use_roi": false, | |
| "beta": 1.0, | |
| "gate": 1.0, | |
| "max_length": 96 | |
| }, | |
| "external_frozen_text_encoder": { | |
| "repo_id": "openai/clip-vit-base-patch32", | |
| "output_dim": 512, | |
| "max_length": 77, | |
| "included": false | |
| }, | |
| "app_vocab": "metadata/app_vocab.json", | |
| "max_accessibility_candidates": 80, | |
| "evaluation": { | |
| "precision": "fp32", | |
| "settings": [ | |
| "AC-Low", | |
| "AC-High" | |
| ], | |
| "history_source": "ground-truth actions", | |
| "subsplit": "IDD", | |
| "episodes": 716, | |
| "steps": 3858, | |
| "seed": 1 | |
| }, | |
| "verified_environment": { | |
| "python": "3.11.14", | |
| "torch": "2.9.0+cu128", | |
| "timm": "1.0.22", | |
| "transformers": "4.57.3" | |
| }, | |
| "code_revision": "c03c565065a8b31f9d4ce1be6c29104c8869799f", | |
| "model_repo_id": "LilCC/GUI-JEPA" | |
| } | |