{"version":1,"type":"story","url":"https://digestai.news/story/finetuned-1-5b-qwen-to-generate-bash-commands-at-gpt-4o-level-using-40","json":"https://digestai.news/story/finetuned-1-5b-qwen-to-generate-bash-commands-at-gpt-4o-level-using-40.json","markdown":"https://digestai.news/story/finetuned-1-5b-qwen-to-generate-bash-commands-at-gpt-4o-level-using-40.md","slug":"finetuned-1-5b-qwen-to-generate-bash-commands-at-gpt-4o-level-using-40","headline":"Finetuned 1.5B Qwen to generate bash commands at gpt-4o level using 400k synthetic examples + Fully opensource finetune dataset","summary":"Finetuned 1.5B Qwen to generate bash commands at gpt-4o level using 400k synthetic examples + Fully opensource finetune dataset\nHow it started\n\nDespite using LLMs for most of the coding, there was always one thing I kept Googling: Bash commands. It's quite flow-breaking to pause work, open Google, type the full query, go to Stack Overflow or similar, and look up the syntax I wanted. Intuitively, this always felt like something a small model would perform well at because you have a finite set of commands, well-defined syntax, and easy-to-generate training data. So one day I decided to actually find out. Over the course of the experiment, I tried six different models: SmolLM 135M, SmolLM 360M, Qwen3-0.6B, Qwen2.5-Coder-1.5B, Qwen3.5-0.8B, and Qwen3.5-2B.","keyPoints":["401,975 synthetic request/command pairs used for fine‑tuning","Model scored 63.7 % on ALFA‑updated 300‑task benchmark"],"whyItMatters":"Fine‑tuned small models that generate shell commands can reduce context‑switching for developers, demonstrating that lightweight LLMs can match larger models on narrow tasks.","category":{"slug":"models","name":"Generative AI & Models","url":"https://digestai.news/category/models"},"entities":{"companies":[],"models":["Qwen3-0.6B","Qwen2.5-Coder-1.5B","Qwen3.5-0.8B","Qwen3.5-2B","SmolLM 135M","SmolLM 360M"],"people":[]},"firstPublishedAt":"2026-10-05T15:38:16Z","updatedAt":"2026-10-05T15:38:16Z","sourceCount":1,"hasPrimarySource":false,"sources":[{"outlet":"dirac.run","title":"Finetuned 1.5B Qwen to generate bash commands at gpt-4o level using 400k synthetic examples + Fully opensource finetune dataset","url":"https://dirac.run/posts/easycommand","publishedAt":"2026-10-05T15:38:16Z","type":"press","primary":false,"lead":true}],"sourceNotes":null,"discussions":[{"site":"Reddit","url":"https://www.reddit.com/r/LocalLLaMA/comments/1wybqq3/finetuned_15b_qwen_to_generate_bash_commands_at/","points":null}],"thread":{"title":"Qwen Bash Command Revolution Unfolds","url":"https://digestai.news/thread/finetuned-1-5b-qwen-to-generate-bash-commands-at-gpt-4o-level-using-400k","storyCount":2},"cite":{"text":"Digest AI, \"Finetuned 1.5B Qwen to generate bash commands at gpt-4o level using 400k synthetic examples + Fully opensource finetune dataset\", 5 October 2026, https://digestai.news/story/finetuned-1-5b-qwen-to-generate-bash-commands-at-gpt-4o-level-using-40","publisher":"Digest AI","title":"Finetuned 1.5B Qwen to generate bash commands at gpt-4o level using 400k synthetic examples + Fully opensource finetune dataset","datePublished":"2026-10-05T15:38:16Z","url":"https://digestai.news/story/finetuned-1-5b-qwen-to-generate-bash-commands-at-gpt-4o-level-using-40"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}