Sam Witteveen Video Watch posted an update
Which is The Best Qwen3.8-27B?
Why it mattersSam Witteveen's description compares ThinkingCap, Swift 1.5 and QwenPi fine-tunes of Qwen3.8-27B, focusing on reducing reasoning tokens while retaining useful accuracy. Chapters include demonstrations and coding, logic and maths results. The compute is Dell-sponsored, and these are the creator's comparisons rather than an independent benchmark audit.
Discuss: How would you test whether a Qwen reasoning fine-tune saves tokens without quietly losing accuracy on your own workload?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.