File "pydantic/main.py", line 341, in pydantic.main.BaseModel.__init__ pydantic.error_wrappers.ValidationError: 1 validation error for LlamaCppEmbeddings

[sandyrs9421]

i am seeing below error when i run the ingest.py. any thoughts on how i can resolve it ? kindly advise

Error -
error loading model: this format is no longer supported (see ggerganov/llama.cpp#1305)
llama_init_from_file: failed to load model
Traceback (most recent call last):
File "/Users/FBT/Desktop/Projects/privategpt/privateGPT/ingest.py", line 39, in
main()
File "/Users/FBT/Desktop/Projects/privategpt/privateGPT/ingest.py", line 30, in main
llama = LlamaCppEmbeddings(model_path="./models/ggml-model-q4_0.bin")
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "pydantic/main.py", line 341, in pydantic.main.BaseModel.init
pydantic.error_wrappers.ValidationError: 1 validation error for LlamaCppEmbeddings
root
Could not load Llama model from path: ./models/ggml-model-q4_0.bin. Received error (type=value_error)

My ENV file -
PERSIST_DIRECTORY=db
MODEL_TYPE=GPT4All
MODEL_PATH=models/ggml-gpt4all-j-v1.3-groovy.bin
EMBEDDINGS_MODEL_NAME=/Users/FBT/Desktop/Projects/privategpt/privateGPT/models/ggml-model-q4_0.bin
MODEL_N_CTX=1000

[christopherpickering]

Did you download and add a model?

[gvilarino]

@sandyRS: as stated in the README.md you should first download the model file into a models directory in the project root. I encountered your same issue and this worked.

[sandyrs9421]

yes i have downloaded the models and mentioned the same path in env file but still seeing the issue.

My ENV file -
PERSIST_DIRECTORY=db
MODEL_TYPE=GPT4All
MODEL_PATH=models/ggml-gpt4all-j-v1.3-groovy.bin
EMBEDDINGS_MODEL_NAME='sentence-transformers/all-MiniLM-L6-v2'
MODEL_N_CTX=1000

can you please help me on how i can resolve thie error ?
@christopherpickering / @gvilarino

[albertas]

Had the same issue. Moving downloaded models into models directory resolved it.

I see that you have MODEL_PATH=models/ggml-gpt4all-j-v1.3-groovy.bin in your .env file, so its likely that the issue is the same for you. Just create models directory and move your downloaded models into it.

[sandyrs9421]

@albertas - thanks for the reply i tried the recommended steps and seeing similar error.
i am seeing this error when i run the privategpt.py file. Data ingestion was successful.
Appreciate if you can guide me wtih possible resolution for this.

My ENV File -
PERSIST_DIRECTORY=db
MODEL_TYPE=GPT4All
MODEL_PATH=models/ggml-gpt4all-j-v1.3-groovy.bin
EMBEDDINGS_MODEL_NAME=all-MiniLM-L6-v2
MODEL_N_CTX=1000
TARGET_SOURCE_CHUNKS=4

[agentmith]

Similar issue, tried with both putting the model in the .\models subfolder and its own folder inside the .\models subdirectory. The only way I can get it to work is by using the originally listed model, which I'd rather not do as I have a 3090. It's most likely a configuration issue in the .env file, but I'm not 100% sure what all to change if you switch to a different model.

[ayteakkaya536]

I had encounter with same problem. I used absolute path for the model path, it resolved the issue.

[Rasmus-Riis]

I had encounter with same problem. I used absolute path for the model path, it resolved the issue.

Thanks. That fixed it for me :-)

[Yousef-Mush]

For me I used absolute path in "privateGPT.py"
previously it was
model_path = os.environ.get('MODEL_PATH')
I changed it to
model_path = "C:/Users/YM/Desktop/PrivateGPT/privateGPT/models/ggml-gpt4all-j-v1.3-groovy.bin"

in .env here is my config:

PERSIST_DIRECTORY=db
MODEL_TYPE=GPT4All
MODEL_PATH=models/ggml-gpt4all-j-v1.3-groovy.bin
EMBEDDINGS_MODEL_NAME=all-MiniLM-L6-v2
MODEL_N_CTX=1000
TARGET_SOURCE_CHUNKS=4

[tosundar40]

Changing the path in the privateGPT.py also does not fix the issue . Please help

[ciathyza]

Any solution found to this yet?
Using absolute path (on MacOS) does not fix it for me.

[abcnow]

same here on mac.. using the absolute path didn't fix the problem

[danielmiranda]

If you set model path correctly mentioned in .env file and remove the extra argument n_ctx=1000, Its working as expected.