Flint – A 30B model fine-tuned for less repetition

Name: Flint – A 30B model fine-tuned for less repetition
Availability: InStock
Author: thmsmxwll

by thmsmxwll·Apr 16, 2026·6 points·2 comments

Visit Project View on HN

AI Analysis

●●SolidNiche GemSolve My Problem

Fine-tuned Qwen 30B that prioritizes output diversity over convergent accuracy.

Strengths

•Targets mode collapse specifically with entropy-focused training rather than generic RLHF.
•Benchmarks claim maintained STEM performance despite divergence tuning focus.
•Addresses the sameness complaint common with current frontier models.

Weaknesses

•Book Demo gatekeeping limits immediate verification of the claims.
•Access restricted to Springboards app, limiting local deployment options.

Post Description

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.

Flint significantly increases the NoveltyBench score compared to the base model, without significantly reducing the score on non-creative benchmarks like MMLU-STEM.

This shows that that divergence tuning doesn't actually have to be a tax on base capabilities.

Flint scores 7.47/10 on NoveltyBench while most frontier models score between 1.8 and 3.2.