Instructions to use coded190/pingti_palm_openvla with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use coded190/pingti_palm_openvla with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("openvla/openvla-7b") model = PeftModel.from_pretrained(base_model, "coded190/pingti_palm_openvla") - Notebooks
- Google Colab
- Kaggle
Download train_config.json from coded190/pingti_palm_openvla: direct link, hf CLI and curl.
- Browser
- Download file 427 Bytes
-
https://huggingface.co/coded190/pingti_palm_openvla/resolve/main/train_config.json
- Command line
-
hf download hf://coded190/pingti_palm_openvla/train_config.json
-
curl -L -o train_config.json https://huggingface.co/coded190/pingti_palm_openvla/resolve/main/train_config.json
427 Bytes
| { | |
| "vla_path": "openvla/openvla-7b", | |
| "data_jsonl": null, | |
| "lerobot_repo_ids": [ | |
| "coded190/pingti_palm_tree", | |
| "coded190/isaac_so_arm101_vla_v2" | |
| ], | |
| "image_root": null, | |
| "output_dir": "outputs/openvla_lora", | |
| "batch_size": 16, | |
| "grad_accum_steps": 1, | |
| "max_steps": 2500, | |
| "save_steps": 200, | |
| "learning_rate": 0.0005, | |
| "mixed_precision": "bf16", | |
| "action_dim": 7, | |
| "predict_stop_token": true, | |
| "seed": 0 | |
| } |