DealForge autonomously sources, scores, and writes investment memos on venture deals. Stop manually hunting.
1,180+ deals tracked · 22 AI investment memos · Updated daily
Show HN: Flint – A 30B model fine-tuned for less repetition
As frontier LLMs have very little output diversity even for open ended queries. We built Flint to see if we could reverse this. It’s a finetuned Qwen3 30B model specifically trained to produce higher entropy when asked open ended questions.<p>Flint significantly increases the NoveltyBench score compared to the base model, without significantly reducing the score on non-creative benchmarks like MMLU-STEM.<p>This shows that that divergence tuning doesn't actually have to be a tax on base capabilities.<p>Flint scores 7.47/10 on NoveltyBench while most frontier models score between 1.8 and 3.2.
Flint addresses a legitimate technical gap in LLMs—the lack of creative diversity—with impressive benchmark results that suggest a high degree of technical proficiency. However, the project currently lacks commercial traction, team visibility, and a clear moat against frontier labs who could implement similar divergence tuning at scale.