Thank you :) Swift Qwen 3.8 27B now has 100k+ downloads, is #1 finetune and #9 model on HuggingFace Trending

The Swift Qwen 3.8 27B model, the first open-source release from its developers, has achieved over 100,000 downloads. It is currently ranked as the #1 finetune and #9 model on HuggingFace Trending. This success demonstrates that penalizing pathological overthinking in small LLMs can reduce token usage by 58.3% and increase speed by 1.95x without sacrificing accuracy, by promoting efficient thinking rather than directly shortening thought processes.