hi,
when we load the model from huggingface:
model = AutoModelForCausalLM.from_pretrained(<model_id>)
How do we know that the model was uploaded after merging the base + adaptor or just the adapter alone is saved ? I merged and uploaded. I see adapter_config.json and adapter_model.safetensors. This does not look different from when the model was uploaded without the merge.
After training with an adapter, one could have merged the base model + adaptor and then uploaded or just uploaded the adapter alone. Can we tell this later when we are loading it ?
Thanks
Mohan