llmfan46

/

gpt-oss-120b-heretic-v2-GGUF

like 9

Thanks for this! How do I bring back the reasoning setting the original gpt-oss-120b has in LM Studio?

You're welcome, but what do you mean by bring back the reasoning setting the original gpt-oss-120b has in LM Studio?

Are you referring to the setting that lmstudio-community has for some of their release where they add a reasoning toggle?

LM Studio has a widget in the prompt box that allows the user to set the reasoning level to be used (low, medium, high). This appears when using the original gpt-oss-20b/120b models, but disappears when using your version.

This appears when using the original gpt-oss-20b/120b models

This one: https://huggingface.co/lmstudio-community/gpt-oss-120b-GGUF

?

Yes, the original unablated one.

Yes, the original unablated one.

That's the lmstudio-community version, this is not the official version and they make their own custom changes like adding a reasoning toggle for when you use it on LM Studio.

My Heretic version is based on the real official version here: https://huggingface.co/openai/gpt-oss-120b

So, there is no way to configure the reasoning strength in your version then?

So, there is no way to configure the reasoning strength in your version then?

Unfortunately not, if the official version does not have it then my version would not have it, the toggles you refer to is a custom job done by the lmstudio-communityspecifically made for when you use the models on LM Studio a toggle is added to make for easy switching like enabling/disabling reasoning and/or setting the reasoning strength (low, medium, high).

But you may be able to enable reasoning manually by adding Reasoning: high (or medium or low) to your system prompt, I dunno if it works though.