River API outperforms Tinker in RL, enabling zero train-infer mismatch for MoEs
marcbhargava · x · 2026-08-18
- Benchmark Result: The River API demonstrated superior performance over Tinker in reinforcement learning runs using identical training code.
- Key Innovation: It successfully achieves zero train-infer mismatch for large Mixture of Experts (MoE) models.
- Optimization Focus: Significant effort was spent on details like routing replay to ensure optimal API results.
- Open Source: The method was validated by teaching Qwen3.6-35B-A3B to play Wordle, with full ablations and open-source code available.
More from Research
- GAE paper learns a geometry-native latent space for 3D-consistent world generation — yshan2u · 2026-10-16
- Agents turn conservative as context grows: edits shrink from 150 to 30 lines — dbreunig · 2026-10-16
- LMR-BENCH at EMNLP 2025 tests whether LLM agents can reproduce real LLM research — ZhiyuChen4 · 2026-10-16
- Ex-OpenAI researcher claims most neurons are legible now; critics call it hype — anshulkundaje · 2026-10-16
- Physicist pushes back on ML community's acceptance that models may be beyond understanding — tszzl · 2026-10-16
- Survey: AI agent evaluation is getting realistic but still fails to measure safety and reliability — burkov · 2026-10-16