Releases: ogre-run/miniogre
v0.11.1
v0.11.0
v0.11.0-beta.1
Partial fix for issue #29
v0.11.0-beta
Fix for issue #26
What changed
Provider ogre is now the default, instead of openai. This enables:
- Any user with a free ogre API Key, obtainable via app.ogre.run, to use miniogre without requiring a OpenAI or similar account
User must add the Ogre API Key to their environment variables:
export OGRE_API_KEY=<token_generated_on_app.ogre.run>
Approach
We want to make it easier for anyone to use miniogre out of the box. Setting ogre as the default provider enables one to do that
pip install miniogre
miniogre run --> this command runs miniogre with the ogre provider
Users can still select other providers (OpenAI, Gemini, etc) by passing miniogre run --provider openai.
IMPORTANT: Services like OpenAI and similar require API Keys generated by premium accounts. The latest models used by miniogre (e.g.
gpt-4ofrom OpenAI) can only be queried if a key from a paying account is present in the environment variables.
v0.10.2
Fixing issue #25
Explanation:
The error you encountered is due to an ast.ImportFrom node having a None value for node.module. This can happen when using relative imports (e.g., from . import data). In this case, node.module is None, so attempting to call .split() on it raises an AttributeError.
Fix:
We need to check if node.module is not None before splitting it.
See file actions.py, line 225
v0.10.1
Fixes and improvements
v0.10.0
Miniogre v0.10.0
- Add
miniogre askto enable users to ask questions about the codebase - Add
miniogre run --provider localfor better local dependency resolution
v0.9.3
Update Gemini model version: flash -> pro.
v0.9.2
Hot fix for dry option which was being overriden.
v0.9.1
New fileserver: https://fileserver.ogrerun.xyz