Executing small model but missing config.json error with microsoft/Phi-3-mini-4k-instruct-gguf about h2ogpt HOT 1 OPEN

Hi, I'll try to clarify the docs, but as in readme.md and in FAQ.md, for new chat template models when using GGUF, you need to specify the tokenizer_base_model.

https://github.com/h2oai/h2ogpt/blob/main/docs/FAQ.md#llama-3-or-other-chat-template-based-models

Also, base_model accepts a few forms of passing urls, TheBloke, etc. for gguf, but not just -gguf, so until I add that, you'll need to use more explicit way of invoking llama.cpp:

e.g. for Phi it would be to use the direct download link like:

python generate.py  --tokenizer_base_model=microsoft/Phi-3-mini-4k-instruct --base_model=llama --llama_cpp_model=https://huggingface.co/microsoft/Phi-3-mini-4k-instruct-gguf/resolve/main/Phi-3-mini-4k-instruct-q4.gguf  --max_seq_len=4096 

or you can download it yourself and place it into llamacpp_path folder.

from h2ogpt.