Skip to content

docs: run the quick start against a local model with Ollama - #241

Merged
fabbrik merged 1 commit into
mainfrom
docs/ollama-quick-start
Nov 17, 2026
Merged

fabbrik merged 1 commit into
mainfrom
docs/ollama-quick-start

Conversation

@fabbrik

@fabbrik fabbrik commented Nov 17, 2026

Copy link
Copy Markdown
Owner

Closes #108.

The sample already supported any OpenAI-compatible endpoint through OPENAI_BASE_URL, so this adds the missing section: commands for Ollama, the model tested (llama3.2:3b), and the real output. In that run the lesson was injected into run 2 and the small model still chose the wrong strategy first. The README says so and points to aexp benefit and the holdout for measuring.

Docs only. QuickStart tests pass (16/16). Two Release.Tests fail locally only because of stale .claude/worktrees directories; they fail the same on a clean main.

🤖 Generated with Claude Code

https://claude.ai/code/session_01GcMT2DEkA15jsvD2sbRRD8

Tested with llama3.2:3b. Records the honest result: the lesson was injected, the small model still picked the wrong strategy first.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GcMT2DEkA15jsvD2sbRRD8
@fabbrik
fabbrik merged commit f3e6549 into main Nov 17, 2026
32 checks passed
@fabbrik
fabbrik deleted the docs/ollama-quick-start branch November 17, 2026 12:57
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Quick start: run the loop against a local model with Ollama

1 participant