invalid model file './models/ggml-alpaca-7b-q4.bin' (too old, regenerate your model files!)

[TonyHanzhiSU]

Hi, I have encounter the above problem when running the alpaca model. I download the model from the link "https://gateway.estuary.tech/gw/ipfs/QmQ1bf2BTnYxq73MFJWu1B7bQ2UD6qG7D7YDCxhTndVkPC" which is one of the three options from the readme. Should I download the model from somewhere else?

[nsarrazin]

You need to convert your weights using the script provided here.

You will also need the tokenizer from here.

Then you can run python convert.py your/models/folder/ path/to/tokenizer.model

And the outputted *.bin.tmp files are the new models.

[nsarrazin]

I have uploaded the converted weights here for the 7B alpaca model. I also have the 13B version but not sure where to upload it, it's a bit big.

[TonyHanzhiSU]

thank you so much ! It has fixed my problem and the model is up and running.

[Zer0-]

I have uploaded the converted weights here for the 7B alpaca model. I also have the 13B version but not sure where to upload it, it's a bit big.

Can you just make a torrent of it for now?

[tkafka]

@nsarrazin You should be able to publish them to https://huggingface.co, right?

[l29ah]

Downloaded one from https://huggingface.co/Pi3141/alpaca-13B-ggml/ and it's "too old" as well.

[l29ah]

magnet:?xt=urn:btih:e5322ab4676e24632a907fd9846234bb40265c4f&dn=ggml-q4 should have the proper models.