Supervised Fine-tuning for Robotic Task Learning
11 repos
Model training and fine-tuning frameworks for robotics, particularly focused on learning manipulation and navigation tasks from demonstrations. The cluster centers on supervised fine-tuning (SFT) approaches applied to benchmark environments like LIBERO and RLBench, using efficient tensor storage formats. Repositories here explore how language models and vision-language models can be adapted to control robotic systems through task-specific training on curated datasets.
safetensors ·8
