Skip to main content

Indie game storeFree gamesFun gamesHorror games
Game developmentAssetsComics
SalesBundles
Jobs
TagsGame Engines

Thank you for all this powerfull softwares ! I have a problem with

LocalGPT Llama2-7b (w/o gui) [CUDA(tokenizing only, chat on cpu)/CPU] 

i launch the first .bat to ingest a document (a pdf in folder SOURCE_DOCUMENTS)., a message shows loading document... and that's all ! console command is not freezed but nothing seems to work, i wait some hours...

so i launch the last .bat to start localgpt, i have errors : 

"

CUDA extension not installed.

CUDA extension not installed.

2024-03-27 10:54:55,908 - INFO - run_localGPT.py:244 - Running on: cuda

2024-03-27 10:54:55,908 - INFO - run_localGPT.py:245 - Display Source Documents set to: False

2024-03-27 10:54:55,909 - INFO - run_localGPT.py:246 - Use history set to: False

2024-03-27 10:54:57,360 - INFO - SentenceTransformer.py:66 - Load pretrained SentenceTransformer: hkunlp/instructor-large

load INSTRUCTOR_Transformer

D:\LocalGPT\miniconda3\lib\site-packages\torch\_utils.py:776: UserWarning: TypedStorage is deprecated. It will be removed in the future and UntypedStorage will be the only storage class. This should only matter to you if you are using storages directly.  To access UntypedStorage directly, use tensor.untyped_storage() instead of tensor.storage()

  return self.fget.__get__(instance, owner)()

max_seq_length  512

2024-03-27 10:55:00,011 - INFO - run_localGPT.py:132 - Loaded embeddings from hkunlp/instructor-large

2024-03-27 10:55:00,260 - INFO - run_localGPT.py:60 - Loading Model: TheBloke/Llama-2-7b-Chat-GGUF, on: cuda

2024-03-27 10:55:00,260 - INFO - run_localGPT.py:61 - This action can take a few minutes!

2024-03-27 10:55:00,262 - INFO - load_models.py:38 - Using Llamacpp for GGUF/GGML quantized models

Traceback (most recent call last):

  File "D:\LocalGPT\localGPT\run_localGPT.py", line 285, in <module>

    main()

  File "D:\LocalGPT\miniconda3\lib\site-packages\click\core.py", line 1157, in __call__

    return self.main(*args, **kwargs)

  File "D:\LocalGPT\miniconda3\lib\site-packages\click\core.py", line 1078, in main

    rv = self.invoke(ctx)

  File "D:\LocalGPT\miniconda3\lib\site-packages\click\core.py", line 1434, in invoke

    return ctx.invoke(self.callback, **ctx.params)

  File "D:\LocalGPT\miniconda3\lib\site-packages\click\core.py", line 783, in invoke

    return __callback(*args, **kwargs)

  File "D:\LocalGPT\localGPT\run_localGPT.py", line 252, in main

    qa = retrieval_qa_pipline(device_type, use_history, promptTemplate_type=model_type)

  File "D:\LocalGPT\localGPT\run_localGPT.py", line 142, in retrieval_qa_pipline

    llm = load_model(device_type, model_id=MODEL_ID, model_basename=MODEL_BASENAME, LOGGING=logging)

  File "D:\LocalGPT\localGPT\run_localGPT.py", line 65, in load_model

    llm = load_quantized_model_gguf_ggml(model_id, model_basename, device_type, LOGGING)

  File "D:\LocalGPT\localGPT\load_models.py", line 56, in load_quantized_model_gguf_ggml

    return LlamaCpp(**kwargs)

  File "D:\LocalGPT\miniconda3\lib\site-packages\langchain\load\serializable.py", line 74, in __init__

    super().__init__(**kwargs)

  File "pydantic\main.py", line 341, in pydantic.main.BaseModel.__init__

pydantic.error_wrappers.ValidationError: 1 validation error for LlamaCpp

__root__

  Could not load Llama model from path: ./models\models--TheBloke--Llama-2-7b-Chat-GGUF\snapshots\191239b3e26b2882fb562ffccdd1cf0f65402adb\llama-2-7b-chat.Q4_K_M.gguf. Received error [WinError -1073741795] Windows Error 0xc000001d (type=value_error)

"

can you help me ? i have 3060 RTX and Xeon E5 with 64 GB RAM?

thank you for all.

localgpt for some reasons dont accepts all documents, on some it stucks and dont works, i dont know how to fix that and yes, this is annoying.

your llama.cpp dont works in 2 bat bcause you are trying to run llama on cuda, when cublas doesnt exist in repack because cpu is enough. To fix that, add set CUDA_VISIBLE_DEVICES=-1 in bat file. Soon ill reupload it with this fix

thank you a lot for your answer. i will try the new version.

When i ingest a document (constitution.pdf from localgpt github) , i don't have error, but it seems to be stucked after the message loaded, the cursor blinked but nothing else...

ingest.py:147 - Loading documents from D:\LocalGPT\localGPT/SOURCE_DOCUMENTS

Importing: constitution.pdf

D:\LocalGPT\localGPT/SOURCE_DOCUMENTS\constitution.pdf loaded.

So, i run the second bat, here i have a error : 

CUDA extension not installed.

CUDA extension not installed.

2024-05-02 06:37:16,651 - INFO - run_localGPT.py:244 - Running on: cpu

2024-05-02 06:37:16,652 - INFO - run_localGPT.py:245 - Display Source Documents set to: False

2024-05-02 06:37:16,652 - INFO - run_localGPT.py:246 - Use history set to: False

2024-05-02 06:37:18,572 - INFO - SentenceTransformer.py:66 - Load pretrained SentenceTransformer: hkunlp/instructor-large

load INSTRUCTOR_Transformer

D:\LocalGPT\miniconda3\lib\site-packages\torch\_utils.py:776: UserWarning: TypedStorage is deprecated. It will be removed in the future and UntypedStorage will be the only storage class. This should only matter to you if you are using storages directly.  To access UntypedStorage directly, use tensor.untyped_storage() instead of tensor.storage()

  return self.fget.__get__(instance, owner)()

max_seq_length  512

2024-05-02 06:37:21,224 - INFO - run_localGPT.py:132 - Loaded embeddings from hkunlp/instructor-large

2024-05-02 06:37:21,655 - INFO - run_localGPT.py:60 - Loading Model: TheBloke/Llama-2-7b-Chat-GGUF, on: cpu

2024-05-02 06:37:21,655 - INFO - run_localGPT.py:61 - This action can take a few minutes!

2024-05-02 06:37:21,656 - INFO - load_models.py:38 - Using Llamacpp for GGUF/GGML quantized models

Traceback (most recent call last):

  File "D:\LocalGPT\localGPT\run_localGPT.py", line 285, in <module>

    main()

  File "D:\LocalGPT\miniconda3\lib\site-packages\click\core.py", line 1157, in __call__

    return self.main(*args, **kwargs)

  File "D:\LocalGPT\miniconda3\lib\site-packages\click\core.py", line 1078, in main

    rv = self.invoke(ctx)

  File "D:\LocalGPT\miniconda3\lib\site-packages\click\core.py", line 1434, in invoke

    return ctx.invoke(self.callback, **ctx.params)

  File "D:\LocalGPT\miniconda3\lib\site-packages\click\core.py", line 783, in invoke

    return __callback(*args, **kwargs)

  File "D:\LocalGPT\localGPT\run_localGPT.py", line 252, in main

    qa = retrieval_qa_pipline(device_type, use_history, promptTemplate_type=model_type)

  File "D:\LocalGPT\localGPT\run_localGPT.py", line 142, in retrieval_qa_pipline

    llm = load_model(device_type, model_id=MODEL_ID, model_basename=MODEL_BASENAME, LOGGING=logging)

  File "D:\LocalGPT\localGPT\run_localGPT.py", line 65, in load_model

    llm = load_quantized_model_gguf_ggml(model_id, model_basename, device_type, LOGGING)

  File "D:\LocalGPT\localGPT\load_models.py", line 56, in load_quantized_model_gguf_ggml

    return LlamaCpp(**kwargs)

  File "D:\LocalGPT\miniconda3\lib\site-packages\langchain\load\serializable.py", line 74, in __init__

    super().__init__(**kwargs)

  File "pydantic\main.py", line 341, in pydantic.main.BaseModel.__init__

pydantic.error_wrappers.ValidationError: 1 validation error for LlamaCpp

__root__

  Could not load Llama model from path: ./models\models--TheBloke--Llama-2-7b-Chat-GGUF\snapshots\191239b3e26b2882fb562ffccdd1cf0f65402adb\llama-2-7b-chat.Q4_K_M.gguf. Received error [WinError -1073741795] Windows Error 0xc000001d (type=value_error)

Please help me, i really want / need this localgpt to work !, thank you for all

Localgpt is buggy and extremely unstable, so: 

1) create small .txt file with some text  and try to ingest, transformating big document into tokens can be REALLY long and i havent tested it cause ima lazy shit

2) try to 2) .bat file and replace paths to llama model into absolute paths (right click on ggml file and click copy path to file and paste this path without quotemarks) 

This may help i think

Hi,  thank you for your answer !

i unzip your localgpt archive to d:\localgpt\

that's ok to ingest a little txt file, it's works !

But it seems i have the same error with i launch 2.bat.

I edit the 2.bat and at the line "set PATH=", i add to the end my path, so i have :

set PATH=C:\Windows\system32;C:\Windows;%DIR%\git\bin;%DIR%;%DIR%\Scripts;D:\LocalGPT\localGPT\models\models--TheBloke--Llama-2-7b-Chat-GGUF\snapshots\191239b3e26b2882fb562ffccdd1cf0f65402adb;

but i have " Could not load Llama model from path: ./models\models--TheBloke--Llama-2-7b-Chat-GGUF\snapshots\191239b3e26b2882fb562ffccdd1cf0f65402adb\llama-2-7b-chat.Q4_K_M.gguf. Received error [WinError -1073741795] Windows Error 0xc000001d (type=value_error)"

i don't understand where is the problem ! i delete the folder models--TheBloke--Llama-2-7b-Chat-GGUF\snapshots\191239b3e26b2882fb562ffccdd1cf0f65402adb\llama-2-7b-chat.Q4_K_M.gguf and 2.bat re download the model (create folders and sub folders correctly) but there is alway the same error when trying to load model !

Please help.

Thank you for all, and sorry to be too noob !

Ummm, try to load it from c: drive, not d:, maybe it ll works but it worked fine in all locations on my pc