Alex Wang

It seems the finetuning result exists. Could you try running `python gradio-chat.py --baseonly` and see if you can successfully run the base model without adapters?

1. AITK uses ONNX runtime GenAI for inference and `frequency_penalty` is converted to `repetition_penalty` behind the scene. 2. According to [ONNX documentation](https://github.com/microsoft/onnxruntime/blob/main/docs/ContribOperators.md), `repetition_penalty` cannot be 0. 3. As the tooltip...

Please check out [this document](https://github.com/microsoft/vscode-ai-toolkit/blob/main/doc/models.md#bring-your-own-models) for adding Ollama models to AI Toolkit. According to this document, the URL format should be like this: > Use "http://localhost:11434/v1/chat/completions" as URL and leave...

@chunlea Do you still have any questions? If not, I will close this issue.

> [@a1exwang](https://github.com/a1exwang) The problem is that when you enter a wrong URL, you can no longer edit anything—at least in Mac OS. It deletes everything and prevents you from getting...

Before this bug is fixed, if anyone encounters this bug, use the following workaround to unblock yourself: 1. Find `my-models.yml` file - Windows: `C:\Users\\.aitk\models\my-models.yml` - macOS: `/Users//.aitk/models/my-models.yml` - Linux: `/home//.aitk/models/my-models.yml`...

The fix will be in the next version. Don't close until next release.

> > Naming of model issue also would be fixed in April's release. > > I had two separate naming problems: > > * In **My Models**, different variants of...

Ollama display name now contains quantization 

Hi @onkanat, after you select `Provide details to add the model` here You can input a custom Ollama endpoint that points to your other machine. The format is something like...