support reasoning effort flag #1087
Conversation
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
| @@ -21,6 +21,7 @@ Options: | |||
| -sm, --small-model <string> 'small' alias model | |||
| -vm, --vision-model <string> 'vision' alias model | |||
| -ma, --model-alias <nameid...> model alias as name=modelid | |||
| -re, --reasoning-effort <string> Reasoning effort for o* models (choices: "high", "medium", "low") | |||
| -lp, --logprobs enable reporting token probabilities | |||
There was a problem hiding this comment.
Missing space between -re and --reasoning-effort.
AI-generated content by pr-docs-review-commit
missing_spacemay be incorrect
|
LGTM 🚀 The pull request adds a new parameter Here's a quick summary:
The changes are clear and well-integrated. There don't appear to be any functional issues or typos in the provided diffs. The new parameter could be beneficial for certain models and provides flexibility in how the application leverages these models. With these additions, the application should be better equipped to handle varying degrees of reasoning effort depending on model capabilities and user needs.
|
|
|
||
| The OpenAI reasoning models, the `o1, o3` models, are models that are optimized for reasoning tasks. | ||
|
|
||
| ```js |
There was a problem hiding this comment.
Invalid code block syntax. Expected language identifier after triple backticks.
AI-generated content by pr-docs-review-commit
invalid_code_blockmay be incorrect
|
|
||
| The reasoning effort parameter can be set to `low`, `medium`, or `high`. | ||
|
|
||
| ```js 'reasoningEffort: "high"' |
There was a problem hiding this comment.
Missing space after comma in option description.
AI-generated content by pr-docs-review-commit
missing_spacemay be incorrect
Summary of Changes
This commit introduces a new reasoning effort parameter to the codebase. Here's a high-level overview of what was changed:
New Property: A new property called
reasoningEfforthas been added to the public API, specifically inpackages/core/src/types/prompt_template.d.ts. This property allows users to specify the level of reasoning effort in their models.Public Facing Changes: The
PromptScriptRunOptionsinterface now includes thereasoningEffortproperty. This addition is user-facing and will allow developers to directly interact with and utilize the new feature.Implementation Updates: The changes also include minor updates throughout the codebase, ensuring that the new
reasoningEffortproperty is properly handled and integrated where necessary.