Datasette LLM 0.1a7 Adds Per‑Model Default Configuration
Explore the new per‑model default configuration in datasette‑llm 0.1a7 to enforce consistent temperature settings across enrichment calls.
Test the new default options in your Datasette LLM plugin to ensure temperature settings apply correctly.
Summary
Datasette has released version 0.1a7 of its datasette‑llm plugin on GitHub. The update introduces a configuration mechanism that lets developers set default options for individual LLM models. By specifying a model in the config, all enrichment operations can now automatically use that model with a preset temperature of 0.5. This change simplifies the workflow for teams that rely on consistent model behaviour across multiple enrichment tasks.
The plugin now supports per‑model defaults, allowing fine‑grained control over parameters such as temperature, top‑p, and max tokens. The new feature is documented in the README under the configuration section, and developers can add entries like `model: my-model, temperature: 0.5` to the config file. Datasette’s LLM ecosystem continues to grow, and this release removes the need to manually pass options for each enrichment call.
Key changes
- Added per‑model default configuration support
- Default options can now be set in a config file
- Temperature can be preset to 0.5 for enrichment operations
- Simplifies consistent model behaviour across tasks
- Updated README with configuration examples