prime-rl 0.9.0 ships adaptive concurrency, online agentic evals during SFT, CPU optimizer offload
samsja19 · x · 2026-08-26
prime-rl 0.9.0 is out, a release its author calls the one he's proudest of — favorite features added while removing a lot of code:
- Adaptive Concurrency
- Online agentic evals during SFT: an inference server runs next to the SFT trainer to execute agentic evals, reusing the weight-broadcast infra to pick up fresh weight updates
- Local run dashboard: panes for logs/metrics, rollout and config viewing
- Full CPU optimizer offload: gradients, optimizer stats and the optimizer step fully offloaded to CPU, with all communication and computation hidden during backward; at long sequence lengths where attention dominates step time, the offloading can be fully hidden
- Custom kernels shipped
More from coding & agent
- Developer Builds Unified Workspace to Debug Multi-Agent LLM Swarms — Impressive-Iron5216 · 2026-08-26
- Sentry founder recommends Cloud or CLI for agent integrations, new release coming — zeeg · 2026-08-26
- Inventory Blindness: Why AI Agents Forget the Skills They Already Have — TechPreacher · 2026-08-26
- Fable struggles with ripgrep flags, adding erroneous -r unlike Codex — andersonbcdefg · 2026-08-26
- Zero-dependency Python tool resolves relative links and inlines images as base64 for multimodal LLMs — Mediocre-Ease4060 · 2026-08-26
- Agent Memory Benchmark Exam Released for First-Person Contexts — LowDistribution3995 · 2026-08-26