slm-lab-benchmark

What problem does it solve?

This Skill automates the process of running deep reinforcement learning benchmarks using the SLM-Lab framework, including monitoring job status, extracting results, and updating benchmark documentation.

Core Features & Use Cases

• Benchmark Execution: Launch and manage deep RL training runs on remote GPUs.
• Result Monitoring: Track job progress and extract key metrics like total reward.
• Documentation Updates: Automatically update benchmark tables with new scores and Hugging Face links, and generate performance plots.
• Use Case: When asked to "run the PPO algorithm on the Pong-v5 environment and report the results," this Skill will initiate the training, monitor its completion, extract the final score, and update the relevant benchmark documentation.

Quick Start

Use the slm-lab-benchmark skill to launch a remote training run for the Pong-v5 environment using the PPO algorithm.