Nicheloom

Market intelligence for builders — see what's gaining traction before it's crowded.

I fine-tuned Qwen 3.5 (0.8B–4B) on a Mac for text-to-SQL

2B beats 12B

Details

External ID
47261691
Source
HN
Company
—
Product
I fine-tuned Qwen 3.5 (0.8B–4B) on a Mac for text-to-SQL
Website domain
github.com
Launched
March 5, 2026
Cohort
—
Upvotes
7
Upvotes percentile
0.4108241082410824
Tags
—
Fetched at
Sept. 7, 2026, 9:26 p.m.
Updated at
Sept. 7, 2026, 9:26 p.m.

Enrichment

Theme
local AI inference and ComfyUI tools
Vertical
Horizontal
Function
Model & infra
Audience
Developer
AI stance
AI-native
Project type
Hobby / open-source project
Normalized one-liner
fine-tuned qwen for text-to-sql
Manually corrected
False

Could you build this?

Yes Fine-tuning open-source models on Apple Silicon using tools like MLX-LM or Unsloth with existing Text-to-SQL datasets is well-documented and easily scripted with an AI assistant.

Discussion

1 comment analyzed.

Competitors mentioned: Mistral-Nemo 12B, Qwen 3.5 Small models

Concerns raised: Larger models hallucinate answers instead of following structured output format, Larger models compute mentally and output plain numbers instead of SQL

Competitors

Other products that read as similar to this one — 267 launches clear the similarity bar, closest 8 shown.

Attention rank: #158 of 268 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).

Launched 125 days after the earliest competitor.

Other launches for this product

Same idea, different domain

Nobody's really built a model & infra tool for Fintech yet.