edit: 임베딩 모델 로컬로 변경

[Gyuho-Han]

임베딩 모델 다시 qwen3-embedding:8b 로 변경

Summary by CodeRabbit

• New Features

  • Switched the app’s default AI/embedding setup to local Ollama models, reducing reliance on external API keys.
  • Updated the recommended LLM example to a newer model version.

• Documentation

  • Revised setup and version notes to reflect the latest model configuration.
  • Documented that existing vector data may need to be regenerated after the embedding change.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 536f001d-245d-4692-bd41-4f1845f0d257

📥 Commits

Reviewing files that changed from the base of the PR and between 25ed014 and 9d1ae88.

⛔ Files ignored due to path filters (1)

• cleaned_dataset.csv is excluded by !**/*.csv

📒 Files selected for processing (6)

• README.md
• VERSIONS.md
• config.py
• ingest_vector_db.py
• requirements.txt
• retriever.py

---

📝 Walkthrough

Walkthrough

Replaces the Google Gemini embedding backend with local Ollama (qwen3-embedding:8b) across config.py, ingest_vector_db.py, and retriever.py. Removes GEMINI_API_KEY, Gemini task-type constants, and the langchain-google-genai dependency. Updates LLM_MODEL to qwen3:14b and documents all changes in VERSIONS.md and README.md.

Changes

Gemini → Ollama Embedding Migration

Layer / File(s)	Summary
Config constants and dependency update config.py, requirements.txt	config.py removes GEMINI_API_KEY, EMBEDDING_TASK_TYPE_DOC/QUERY, sets EMBEDDING_MODEL to qwen3-embedding:8b, adds OLLAMA_BASE_URL, and updates LLM_MODEL to qwen3:14b. requirements.txt drops langchain-google-genai>=2.0.
OllamaEmbeddings wiring in ingest and retriever ingest_vector_db.py, retriever.py	Both files replace GoogleGenerativeAIEmbeddings construction with OllamaEmbeddings using EMBEDDING_MODEL and OLLAMA_BASE_URL, updating imports and initialization logic accordingly.
Documentation updates README.md, VERSIONS.md	README updates the required LLM model example to qwen3:14b. VERSIONS.md adds v0.0.8 (Ollama embedding switch) and v0.0.7 (LLM upgrade) entries, and relabels v0.0.6 as a previous version.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Possibly related PRs

• Gyuho-Han/chatdds#8: Introduced Google Gemini embeddings and GEMINI_API_KEY into the same config.py, ingest_vector_db.py, and retriever.py paths that this PR reverts to Ollama.

Poem

🐇 Hop hop, no more Gemini key,
Local Ollama runs wild and free!
qwen3 embeddings, fresh and bright,
chroma_db reborn overnight.
The rabbit cheers — all local, all right! 🎉

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch v0.0.6

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}