Discussions

hello, i wrote this in huggingface in your model discussion too, in hugginface, your models are not the actual model, they are adapter models, they are only weights of fine tuned part of the actual model. and you cant directly use it because its a part of model. you have to merge main model with your adapter model. i had the same issue too and i can help you merge your adapter weights with main model. you can check my model in my profile, which is a fine tuned mistral-7b-v0.2 which is fine tuned by me with these steps : fine tuned using qlora, merged adapter weights with main model and quantized it in gguf. you can contact me from email or hugginface.