CoolFace
Modelpublic

devsheroubi/xvla-stackcupsv6

sourceHugging Faceapache-2.0updated 9mo agoView on Hugging Face
0likes8downloads
train_config.json355 linesDownload Raw Back to root
1{2    "dataset": {3        "repo_id": "devsheroubi/stackcupsv4",4        "root": null,5        "episodes": null,6        "image_transforms": {7            "enable": false,8            "max_num_transforms": 3,9            "random_order": false,10            "tfs": {11                "brightness": {12                    "weight": 1.0,13                    "type": "ColorJitter",14                    "kwargs": {15                        "brightness": [16                            0.8,17                            1.218                        ]19                    }20                },21                "contrast": {22                    "weight": 1.0,23                    "type": "ColorJitter",24                    "kwargs": {25                        "contrast": [26                            0.8,27                            1.228                        ]29                    }30                },31                "saturation": {32                    "weight": 1.0,33                    "type": "ColorJitter",34                    "kwargs": {35                        "saturation": [36                            0.5,37                            1.538                        ]39                    }40                },41                "hue": {42                    "weight": 1.0,43                    "type": "ColorJitter",44                    "kwargs": {45                        "hue": [46                            -0.05,47                            0.0548                        ]49                    }50                },51                "sharpness": {52                    "weight": 1.0,53                    "type": "SharpnessJitter",54                    "kwargs": {55                        "sharpness": [56                            0.5,57                            1.558                        ]59                    }60                },61                "affine": {62                    "weight": 1.0,63                    "type": "RandomAffine",64                    "kwargs": {65                        "degrees": [66                            -5.0,67                            5.068                        ],69                        "translate": [70                            0.05,71                            0.0572                        ]73                    }74                }75            }76        },77        "revision": null,78        "use_imagenet_stats": true,79        "video_backend": "torchcodec",80        "streaming": false81    },82    "env": null,83    "policy": {84        "type": "xvla",85        "n_obs_steps": 1,86        "input_features": {87            "observation.images.image": {88                "type": "VISUAL",89                "shape": [90                    3,91                    256,92                    25693                ]94            },95            "observation.images.image2": {96                "type": "VISUAL",97                "shape": [98                    3,99                    256,100                    256101                ]102            },103            "observation.state": {104                "type": "STATE",105                "shape": [106                    8107                ]108            },109            "observation.images.empty_camera_0": {110                "type": "VISUAL",111                "shape": [112                    3,113                    224,114                    224115                ]116            }117        },118        "output_features": {119            "action": {120                "type": "ACTION",121                "shape": [122                    6123                ]124            }125        },126        "device": "cuda",127        "use_amp": false,128        "push_to_hub": true,129        "repo_id": "devsheroubi/xvla-stackcupsv6",130        "private": null,131        "tags": null,132        "license": null,133        "pretrained_path": "lerobot/xvla-base",134        "chunk_size": 30,135        "n_action_steps": 30,136        "dtype": "bfloat16",137        "normalization_mapping": {138            "STATE": "IDENTITY",139            "ACTION": "MEAN_STD",140            "VISUAL": "IDENTITY"141        },142        "florence_config": {143            "model_type": "florence2",144            "bos_token_id": 0,145            "eos_token_id": 2,146            "ignore_index": -100,147            "pad_token_id": 1,148            "projection_dim": 1024,149            "text_config": {150                "vocab_size": 51289,151                "activation_dropout": 0.1,152                "activation_function": "gelu",153                "add_bias_logits": false,154                "add_final_layer_norm": false,155                "attention_dropout": 0.1,156                "bos_token_id": 0,157                "classif_dropout": 0.1,158                "classifier_dropout": 0.0,159                "d_model": 1024,160                "decoder_attention_heads": 16,161                "decoder_ffn_dim": 4096,162                "decoder_layerdrop": 0.0,163                "decoder_layers": 12,164                "decoder_start_token_id": 2,165                "dropout": 0.1,166                "early_stopping": true,167                "encoder_attention_heads": 16,168                "encoder_ffn_dim": 4096,169                "encoder_layerdrop": 0.0,170                "encoder_layers": 12,171                "eos_token_id": 2,172                "forced_eos_token_id": 2,173                "forced_bos_token_id": 0,174                "gradient_checkpointing": false,175                "init_std": 0.02,176                "is_encoder_decoder": true,177                "label2id": {178                    "LABEL_0": 0,179                    "LABEL_1": 1,180                    "LABEL_2": 2181                },182                "max_position_embeddings": 4096,183                "no_repeat_ngram_size": 3,184                "normalize_before": false,185                "num_hidden_layers": 12,186                "pad_token_id": 1,187                "scale_embedding": false,188                "num_beams": 3189            },190            "vision_config": {191                "model_type": "davit",192                "drop_path_rate": 0.1,193                "patch_size": [194                    7,195                    3,196                    3,197                    3198                ],199                "patch_stride": [200                    4,201                    2,202                    2,203                    2204                ],205                "patch_padding": [206                    3,207                    1,208                    1,209                    1210                ],211                "patch_prenorm": [212                    false,213                    true,214                    true,215                    true216                ],217                "enable_checkpoint": false,218                "dim_embed": [219                    256,220                    512,221                    1024,222                    2048223                ],224                "num_heads": [225                    8,226                    16,227                    32,228                    64229                ],230                "num_groups": [231                    8,232                    16,233                    32,234                    64235                ],236                "depths": [237                    1,238                    1,239                    9,240                    1241                ],242                "window_size": 12,243                "projection_dim": 1024,244                "visual_temporal_embedding": {245                    "type": "COSINE",246                    "max_temporal_embeddings": 100247                },248                "image_pos_embed": {249                    "type": "learned_abs_2d",250                    "max_pos_embeddings": 50251                },252                "image_feature_source": [253                    "spatial_avg_pool",254                    "temporal_avg_pool"255                ]256            },257            "vocab_size": 51289,258            "torch_dtype": "float32",259            "is_encoder_decoder": true260        },261        "tokenizer_name": "facebook/bart-large",262        "tokenizer_max_length": 1024,263        "tokenizer_padding_side": "right",264        "pad_language_to": "max_length",265        "hidden_size": 1024,266        "depth": 24,267        "num_heads": 16,268        "mlp_ratio": 4.0,269        "num_domains": 30,270        "len_soft_prompts": 32,271        "dim_time": 32,272        "max_len_seq": 512,273        "use_hetero_proj": false,274        "action_mode": "auto",275        "num_denoising_steps": 10,276        "use_proprio": true,277        "max_state_dim": 20,278        "max_action_dim": 20,279        "domain_feature_key": null,280        "resize_imgs_with_padding": [281            224,282            224283        ],284        "num_image_views": 4,285        "empty_cameras": 1,286        "freeze_vision_encoder": false,287        "freeze_language_encoder": false,288        "train_policy_transformer": true,289        "train_soft_prompts": true,290        "optimizer_lr": 0.0001,291        "optimizer_betas": [292            0.9,293            0.95294        ],295        "optimizer_eps": 1e-08,296        "optimizer_weight_decay": 0.0001,297        "optimizer_grad_clip_norm": 10.0,298        "optimizer_soft_prompt_lr_scale": 1.0,299        "optimizer_soft_prompt_warmup_lr_scale": null,300        "scheduler_warmup_steps": 1000,301        "scheduler_decay_steps": 30000,302        "scheduler_decay_lr": 2.5e-06303    },304    "output_dir": "outputs/train/stackcupsv6",305    "job_name": "stackcupsv4",306    "resume": false,307    "seed": 1000,308    "num_workers": 4,309    "batch_size": 32,310    "steps": 20000,311    "eval_freq": 20000,312    "log_freq": 200,313    "save_checkpoint": true,314    "save_freq": 5000,315    "use_policy_training_preset": true,316    "optimizer": {317        "type": "xvla-adamw",318        "lr": 0.0001,319        "weight_decay": 0.0001,320        "grad_clip_norm": 10.0,321        "betas": [322            0.9,323            0.95324        ],325        "eps": 1e-08,326        "soft_prompt_lr_scale": 1.0,327        "soft_prompt_warmup_lr_scale": null328    },329    "scheduler": {330        "type": "cosine_decay_with_warmup",331        "num_warmup_steps": 1000,332        "num_decay_steps": 30000,333        "peak_lr": 0.0001,334        "decay_lr": 2.5e-06335    },336    "eval": {337        "n_episodes": 50,338        "batch_size": 50,339        "use_async_envs": false340    },341    "wandb": {342        "enable": true,343        "disable_artifact": false,344        "project": "lerobot",345        "entity": null,346        "notes": null,347        "run_id": "o0q2m0fe",348        "mode": null349    },350    "checkpoint_path": null,351    "rename_map": {352        "observation.images.mount": "observation.images.image",353        "observation.images.front": "observation.images.image2"354    }355}