ggml-alpaca-7b-q4.bin: invalid model file (bad magic [got 0x67676d66 want 0x67676a74])

[ShoufaChen]

When loading the converted ggml-alpaca-7b-q4.bin model, I met the error:

>>> llm = LlamaCpp(model_path="ggml-alpaca-7b-q4.bin")
llama_model_load: loading model from 'ggml-alpaca-7b-q4.bin' - please wait ...
ggml-alpaca-7b-q4.bin: invalid model file (bad magic [got 0x67676d66 want 0x67676a74])
        you most likely need to regenerate your ggml files
        the benefit is you'll get 10-100x faster load times
        see https://github.com/ggerganov/llama.cpp/issues/91
        use convert-pth-to-ggml.py to regenerate from original pth
        use migrate-ggml-2023-03-30-pr613.py if you deleted originals
llama_init_from_file: failed to load model

[abetlen]

Hi @ShoufaChen unfortunately this has to do with a recent change to the model format in llama.cpp. To fix this you'll just need to migrate the model flie as follows.

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
python3 migrate-ggml-2023-03-30-pr613.py ggml-alpaca-7b-q4.bin ggml-alpaca-7b-q4-new.bin

[ShoufaChen]

Hi @abetlen

Thanks for your help. I believe #2393 could help.

[akash-ravikumar]

migrate-ggml-2023-03-30-pr613.py does not seem to exist on this repo: https://github.com/ggerganov/llama.cpp.git

[jav-ed]

I also can't find it, instead convert-lora-to-ggml.py, convert-pth-to-ggml.py and convert.py are found

[Diyago]

doest start converting

> PS C:\Users\dex\Desktop\gpt4free\voice_chatbot\migrate_models\llama.cpp> python3 convert.py ggml-vic7b-uncensored-q5_1.bin ggml-vic7b-uncensored-q5_1_new.bin

Python

[am0oma]

python convert.py x.bin --outfile x_new.bin
or
python3 convert.py x.bin --outfile x_new.bin

doest start converting

> PS C:\Users\dex\Desktop\gpt4free\voice_chatbot\migrate_models\llama.cpp> python3 convert.py ggml-vic7b-uncensored-q5_1.bin ggml-vic7b-uncensored-q5_1_new.bin

Python

[AlekzNet]

I'm having the same issue. Converting did not help.

$ python3 ./convert.py --outfile models/7B/ggml-model-q4_0_new.bin models/7B/ggml-model-q4_0.bin
Loading model file models/7B/ggml-model-q4_0.bin
Writing vocab...
. . .
Wrote models/7B/ggml-model-q4_0_new.bin

$ ./main -m models/7B/ggml-model-q4_0_new.bin
main: seed = 1686525413
llama_model_load: loading model from 'models/7B/ggml-model-q4_0_new.bin' - please wait ...
llama_model_load: invalid model file 'models/7B/ggml-model-q4_0_new.bin' (bad magic)
main: failed to load model from 'models/7B/ggml-model-q4_0_new.bin'

[vinitran]

i met the same bug like @AlekzNet , anyone has solution ? i've run convert.py successfully then i run ./main -m models/7B/ggml-model-q4_0_new.bin but it show bad magic

[dosubot]

Hi, @ShoufaChen. I'm Dosu, and I'm helping the LangChain team manage their backlog. I wanted to let you know that we are marking this issue as stale.

Based on my understanding of the issue, you reported that the ggml-alpaca-7b-q4.bin model file is invalid and cannot be loaded. There have been suggestions to regenerate the ggml files using the convert-pth-to-ggml.py script or the migrate-ggml-2023-03-30-pr613.py script. However, there are mixed reactions to their effectiveness.

Before we proceed, we would like to confirm if this issue is still relevant to the latest version of the LangChain repository. If it is, please let us know by commenting on this issue. Otherwise, feel free to close the issue yourself, or the issue will be automatically closed in 7 days.

Thank you for your understanding and contribution to the LangChain project. We appreciate your patience and look forward to hearing from you soon.

[dosubot]

Hi, @ShoufaChen. I'm Dosu, and I'm helping the LangChain team manage their backlog. I wanted to let you know that we are marking this issue as stale.

Based on my understanding of the issue, you reported that the ggml-alpaca-7b-q4.bin model file is invalid and cannot be loaded. There have been suggestions to regenerate the ggml files using the convert-pth-to-ggml.py script or the migrate-ggml-2023-03-30-pr613.py script. However, there are mixed reactions to their effectiveness.

Before we proceed, we would like to confirm if this issue is still relevant to the latest version of the LangChain repository. If it is, please let us know by commenting on this issue. Otherwise, feel free to close the issue yourself, or the issue will be automatically closed in 7 days.

Thank you for your understanding and contribution to the LangChain project. We appreciate your patience and look forward to hearing from you soon.