In the news
Training SWE agents with Miles on MI355X: taming long-tail trajectories and complex sandbox coordination
LMSYS · Published · 3 min read
In 30 seconds
- What happened
- Researchers trained a 30B coding model using reinforcement learning on AMD MI355X GPUs, improving task solve rate from 7.3% to 27.5% on SWE-bench.
- Why it matters
- Engineers building agentic AI systems need to understand how to scale RL training for code generation across multiple GPUs and manage sandbox infrastructure efficiently.
- Watch out
- Long-tail trajectories and CPU sandbox bottlenecks dominated training time, not GPU compute. Infrastructure limits like Docker network caps and connection pools broke before network bandwidth became limiting.
- agent
The patterns behind this
Each one covers how the technique works, when it earns its cost, and where it breaks.
The Agent Architect
One pattern, one tradeoff, one production failure story. A short weekly briefing for people building agentic systems.
Weekly email, one-click unsubscribe. We only use your address to send the briefing.