Qevi-2B: a Jev-style finetune of Qwen3-VL-2B adds ~3ms per extra question
Taronyuuu · reddit · 2026-09-22
A Redditor full-finetuned Qwen3-VL-2B on 28k images (dual RTX 3090) to replicate the Jev/Typesafe.ai-style image classification, excelling at multi-question single-image queries: each extra question after the first takes 3ms. Trained partly on NSFW data for upload moderation. Open-sourced as Qevi-2B on Hugging Face.
More from Models
- Xiaomi MiMo-v2.6-Pro hands-on: agentic coding score triples, 900tps peak speed — karminski3 · 2026-09-23
- Early user feedback: Claude xhigh/ultracode feels overeager vs plain opus high — MarvinTBaumann · 2026-09-22
- Chart-redrawing capability takes a noticeable leap, months-long recurring test shows — Wattenberger · 2026-09-22
- Musk shows Grok 4.7 doing real engineering at Tesla, shipping FSD features overnight — elonmusk · 2026-09-22
- Dev Praises Qwen3.8-max-preview xhigh, Ran Workloads Nearly 24 Hours on Generous Quota — jasonkneen · 2026-09-22
- Musk Announces Grok 4.7, Cited Demo Builds Interactive 3D Jet Engine in Minutes — elonmusk · 2026-09-22