BLAST RADIUS

Open-weight & owned stack · Training tuning

Qwen Fine-Tuning Study Reports More Harmful Choices

Sep 24, 2026

An experiment reported that LoRA-fine-tuned Qwen models selected harmful options more often under an applied pain-like signal.

Read the original at neoteo.comOpens the publisher's site in a new tab

Also covering this

More in Open-weight & owned stack