Few Shot Tune Check
LiveTest few shot prompts on your examples and learn if tuning is still needed.
musemuse-few-shot-tune-check.vercel.appOpen (opens in a new tab)
Get it
Connect it to any app that speaks MCP. It has 1 tool: evaluate_few_shot_fit. No sign-in.
- Muse app: add this URL as a custom connector.
- Claude Code:
claude mcp add --transport http few-shot-tune-check https://muse-muse.vercel.app/api/mcp/few-shot-tune-check - Anything else: add it as a remote server over streamable HTTP.
What was checked
Checked by code
Each line is what the check ran on, and no more.
- Passed Typecheck · Oct 28, 2026
- Passed Build · Oct 28, 2026
- Passed The built app starts · Oct 28, 2026
- Passed No errors in the browser console · Oct 28, 2026
- Passed Smoke test of the main flow · Oct 28, 2026
- Passed Its manifest is valid · Oct 28, 2026
Judged by a model
This is a model's opinion, not proof.
- Passed A model's opinion of the screenshots, against the brief · Oct 28, 2026
Not checked
Nothing here should be taken as tested.
- Not checked Whether its answers are true
- Not checked ChatGPT: only a person with an account can test it
- Not checked Installing the Claude plugin or the Muse gadget pack on a real device
- Not checked No Prove it run: nothing it comes with was run for real
The muse
Rebuild of Meta's videoLlama Best Practices: Fine-Tuning Llama Models in the Real World (opens in a new tab)Reframe
Rebuild of a Meta developer video on fine tuning strategy, reframed as a few shot check with hosted models.
Took from it
- choosing when a domain classification task needs fine tuning
Left out
- video footage
- video script
- Meta and Llama names and logos