dify icon indicating copy to clipboard operation
dify copied to clipboard

Feature: Add More LLM Request Parameter Settings

Open wojiaoyishang opened this issue 11 months ago • 1 comments

Self Checks

  • [x] I have searched for existing issues search for existing issues, including closed ones.
  • [x] I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
  • [x] [FOR CHINESE USERS] 请务必使用英文提交 Issue,否则会被关闭。谢谢!:)
  • [x] Please do not modify this template :) and fill in all the required fields.

1. Is this request related to a challenge you're experiencing? Tell me about your story.

Different language models have various adjustable request parameters. Although the LLM generation node comes pre-configured with parameters such as Top-P and temperature, there are additional parameters that vary from different vendors to individual models. For example, the OpenAI documentation also provides adjustments for parameters like reasoning_effort and web_search_options.

Image

In third-party interfaces (such as the qwen2.5-vl model), there is a vl_high_resolution_images parameter to control whether images need to be resized.

Image

2. Additional context or comments

Here, I sincerely propose this feature, hoping that custom parameter settings can be added in the next version of dify. Since this is aimed at advanced users, a simple addition could allow users to pass a JSON-formatted data to include extra parameters.

Thanks!

3. Can you help us with this feature?

  • [x] I am interested in contributing to this feature.

wojiaoyishang avatar Mar 12 '25 09:03 wojiaoyishang

Add: my idea is that all node settings can pass in a global context variable to improve the freedom of node settings, such as the clarity of image recognition can also be replaced by a global context variable.

For example, the node setting freedom of comfyui.

wojiaoyishang avatar Mar 13 '25 11:03 wojiaoyishang

Hi, @wojiaoyishang. I'm Dosu, and I'm helping the Dify team manage their backlog. I'm marking this issue as stale.

Issue Summary

  • You requested customizable request parameter settings for language models in Dify.
  • Suggested using JSON-formatted data for parameters like reasoning_effort and vl_high_resolution_images.
  • Proposed using a global context variable for node settings, similar to comfyui.
  • Expressed willingness to contribute to the development of this feature.

Next Steps

  • Please let me know if this issue is still relevant to the latest version of the Dify repository by commenting here.
  • If there is no further activity, the issue will be automatically closed in 15 days.

Thank you for your understanding and contribution!

dosubot[bot] avatar Apr 13 '25 16:04 dosubot[bot]