Open-weight & owned stack · Training tuning

Community Releases Qwen Humanlike Chat Fine-Tune

Sep 11, 2026

LessThanThreeAI released a Qwen3.8-27B LoRA fine-tune trained on obfuscated human messages for more natural conversation.

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

More in Open-weight & owned stack

Z.ai publishes GLM-5.3 open weightsSep 8 MiniMax H3 highlighted as open-weight multimodal modelSep 7 InternLM releases 397B multimodal Intern-S2 modelSep 13 DeepSeek Model Nearly Matches GPT-6 Astra DesignSep 10 DeepSeek presents V4 FlashSep 8