Update llama-cpp-python from 0.1.45 to 0.1.50

[feeelX]

The ggml format was updated again so newly created models might require a newer llama.cpp version.

This should fix the following open issues:

• llama.cpp: loading model from models\Chinese-LLaMA-7B\ggml-model-q4_0.bin error loading model: unknown (magic, version) combination: 67676a74, 00000002; is this really a GGML file? llama_init_from_file: failed to load model #2020
• is this really a GGML file? llama_init_from_file: failed to load model #2046
• Says, "Is Typing..." But doesn't and resets #2000

[ghost]

You have a double h in "hhttps://github..." (line 21), please fix.

But honestly though we should get this fix in sooner rather than later as llama.cpp upstream as well as many of the GGML HF repos have thoroughly embraced this change. I've been using the UI with v0.1.50 and it's working fine for me.

[feeelX]

You have a double h in "hhttps://github..." (line 21), please fix.

But honestly though we should get this fix in sooner rather than later as llama.cpp upstream as well as many of the GGML HF repos have thoroughly embraced this change. I've been using the UI with v0.1.50 and it's working fine for me.

My bad. Fixed it now.

[oobabooga]

I had to reconvert my llama 7b weights to be able to use this new version.