Expand Ollama documentation (#32)

* Expand Ollama documentation * Separate configuration options and environment variables. Remove table.
gramps-project · Dec 24, 2024 · cd0ec92 · cd0ec92
1 parent d897894
commit cd0ec92
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 1 deletion.
diff --git a/docs/install_setup/chat.md b/docs/install_setup/chat.md
@@ -90,7 +90,9 @@ volumes:
     ollama_data:
 ```
 
-and then set the `LLM_BASE_URL` configuration parameter to `http://ollama:11434/v1` and the `LLM_MODEL` to a model supported by Ollama.
+and then set the `LLM_BASE_URL` configuration parameter to `http://ollama:11434/v1`. Set `LLM_MODEL` to a model supported by Ollama, and pull it down in your container with `ollama pull <model>`.  Finally, set `OPENAI_API_KEY` to `ollama`.
+
+To troubleshoot problems with Ollama, you can enable debug logging by setting environment variable `OLLAMA_DEBUG=1` in the Ollama service environment.
 
 !!! info
     If you are using Ollama for Gramps Web AI chat, please support the community by completing this documentation with any missing details.

diff --git a/docs/install_setup/configuration.md b/docs/install_setup/configuration.md
@@ -96,6 +96,7 @@ Key | Description
 `LLM_BASE_URL` | Base URL for the OpenAI-compatible chat API. Defaults to `None`, which uses the OpenAI API.
 `LLM_MODEL` | The model to use for the OpenAI-compatible chat API. If unset (the default), chat is disabled.
 `VECTOR_EMBEDDING_MODEL` | The [Sentence Transformers](https://sbert.net/) model to use for semantic search vector embeddings. If unset (the default), semantic search and chat are disabled.
+`LLM_MAX_CONTEXT_LENGTH` | Character limit for the family tree context provided to the LLM. Defaults to 50000.
 
 
 ## Example configuration file