Instructions to use IDEAS-Lab-Northwestern/gr00t-sim-table-sft with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use IDEAS-Lab-Northwestern/gr00t-sim-table-sft with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("IDEAS-Lab-Northwestern/gr00t-sim-table-sft", dtype="auto") - LeRobot
How to use IDEAS-Lab-Northwestern/gr00t-sim-table-sft with LeRobot:
- Notebooks
- Google Colab
- Kaggle
GR00T N1.7 SFT on sim-table
This repository stores Isaac-GR00T checkpoints for the sim_table_gr00t
fine-tuning run on IDEAS-Lab-Northwestern/sim-table.
The checkpoint layout mirrors the local training output so future steps can be appended directly:
checkpoints/
sim_table_gr00t/
checkpoint-5000/
config.json
model-*.safetensors
model.safetensors.index.json
processor_config.json
statistics.json
trainer_state.json
...
The dataset was converted with:
scripts/sim_table/convert_sim_table_to_gr00t.py
examples/sim_table/sim_table_config.py
The EEF action group is delta XYZ + delta rotvec in the dataset already, so the modality config avoids GR00T's internal absolute-to-relative action conversion.