RLFR Cuts Hallucinations by 37% with Internal Probes
burny_tech · x · 2026-07-15
The Goodfire team developed RLFR, a method that uses a model's internal probes as reinforcement learning reward signals during training.
They noted that Silico replicated this method in just 2 days, reducing the hallucination rate in Qwen3-8B by 37% without any degradation in the model's capabilities.
Related event: RLFR Method Uses Internal Probes to Reduce LLM Hallucinations by 37%(3 posts)→
More from Models
- Unreleased 'GPT 6 Sol' model spotted in OpenAI's API — SteveEricJordan · 2026-09-11
- DeepSeek update keeps cache hits mid-conversation, cuts costs 36.6% — teortaxesTex · 2026-09-11
- AI Sextet offers 6 models free and unlimited for 14 days, including DeepSeek and Qwen — airesearch12 · 2026-09-11
- BullshitBench update: GPT-6-Astra beats all prior OpenAI models but still trails Anthropic — scaling01 · 2026-09-11
- Astra Scores 83% on GauntletBench, First Computer-Use Agent to Beat Human Baseline — ducha_aiki · 2026-09-11
- Kimi K2.8 Preview rolls out: near-K3 coding performance, 1M context for all tiers — teortaxesTex · 2026-09-11