How to install and use PrivateGPT - Privately Chat with your own documents
ฝัง
- เผยแพร่เมื่อ 18 ก.ย. 2024
- PrivateGPT is a really useful new project that you’ll find really useful.
You can basically load your private text files, PDF documents, powerpoint and use the power of Ai to interact with those documents. And it’s private and on your own computer.
It doesn’t have access to the internet and runs locally on your computer.
I’ll show you exactly how to set it up and use it and I’ll simplify it, so you can do this, even if you don’t know anything about computer programming.
First, you need a program called Visual Studio. You can download that here:
code.visualstu...
Then go to file, new window. Then, make sure you can see the terminal section on the bottom of the page.
Next, you need the code from Github. You can copy the URL of the Github repo here.
github.com/ima...
Go back to Visual Studio. Type cd desktop to access your computer desktop.
Then type: git clone github.com/ima...
That should take a few seconds to install.
Then, go inside that folder by typing cd PrivateGPT
Then type pip install -r requirements.txt
The install should take a few minutes. If you have issues, you can type pip3 to use a specific version of python. If you don’t have python installed, you’ll need to install that first.
www.python.org...
Back in Github, download the LLM model.
Next, look for example.env on the left side of visual studio. And rename it .env.
Now, you can make a new folder called LLM models and move the downloads to it. You can use the finder window to make it easier.
You should see a folder called source-documents. This is where you will add your private documents. There is a test file there by default you can practice with.
Finally, type python ingest.py
This will ingest the documents. If you have an issue, try python3 ingest.py
You can check the Python version in your terminal or command prompt by using the following command: python --version
Or, for Python 3 specifically:python3 --version
To use it, use this command. python privateGPT.py
(Or python3 privateGPT.py)
The answer typically takes 20-30 seconds.
Remember to run the ingest prompt when you add new files to your folder.
Learning ChatGPT, Midjourney and AI just got way easier. Check out the ultimate AI learning platform. bit.ly/skillle...
Learning ChatGPT, Midjourney and AI just got way easier. Check out the ultimate AI learning platform. bit.ly/skillleapai
What CPU you use, how many Memory we need ? my friend need up to 1 minute for one answer.
@@rogerkovac3732 i was expecting that.
@@rogerkovac3732
Operating System: Windows 10/11, macOS, Linux (64-bit)
Processor: 6-core or 8-core CPU (Intel i7 or Ryzen 7 equivalent)
RAM: 32 GB (minimum 16 GB)
Storage: 100 GB free space, preferably on an SSD
GPU: NVIDIA GPU with at least 8GB VRAM (if using GPU)
Python: Version 3.7 or higher
the requirements.txt is no longer in the ZIP... :( What should i do know?
for that you have to pip install poetry -> poetry run python scripts/setup
hey thank you, but it didn't worked for me, when I add the command line "pip install poetry"
@@akki_the_tecki
@@akki_the_tecki poetry run python scripts/setup
poetry : The term 'poetry' is not recognized as the name of a cmdlet, function, script file, or operable program. Check the spelling of the name, or if
a path was included, verify that the path is correct and try again.
At line:1 char:1
+ poetry run python scripts/setup
+ ~~~~~~
+ CategoryInfo : ObjectNotFound: (poetry:String) [], CommandNotFoundException
+ FullyQualifiedErrorId : CommandNotFoundException
@@akki_the_tecki python -m poetry run python scripts/setup
Poetry could not find a pyproject.toml file in "path" or its parents
I suspect this is an old version. I read the docs and it seems the project now uses poetry and not pip so no requirements anymore. Instead there is the pyproject.toml
PS C:\Users\User\Desktop\privateGPT> pip install -r requirements.txt
ERROR: Could not open requirements file: [Errno 2] No such file or directory: 'requirements.txt'
How easy installation in TH-cam BUT in real life its hard as hell
The struggle is real. But definitely easier using WSL instead of bare Windows. But still there are some dependency errors to fix, it's clearly (as stated in README) not production ready... :D
I developed and implemented a web-based version several months ago. Always amazed by all the changes we are experiencing. Where is your env?
Looking at doing the same. Do you mine to share your source code? If not that's cool I can just code it lol
Definitely not beginner friendly. I'm getting all sorts of errors. 😂
Sounds like me😂😂😂
Yo, I closed up shop for today, 6 hours sunk and still errors!
You need to set up the VSC first. Just use another tutorial for VSC
Hello, I am stuck at the "Enter a query" prompt, receiving this error message: "Index not found, please create an instance before querying". Can you please help?
Is this a tutorial about an old version of privateGPT? because not only the installation with pip is not supported anymore but now there are makefiles to run things it seems
[sarcasm]Very user friendly, everybody can install and use it, my wife, my kids, grandma, anyone, even my neighbors.[/sarcasm]
the requirements.txt file is no longer available
Could you find any solution yet?
@@mohorbanerjee5457 no
\privateGPT> pip install -r requirements.txt
ERROR: Could not open requirements file: [Errno 2] No such file or directory: 'requirements.txt'
it looks like the 'requirements.txt' is not there.
Any thoughts on how to get past this?
I suspect that now the system uses poetry not pip and therefore no requirements anymore. Google poetry (python) to see how to install dependencies using poetry
this link don't match to what he has shown and there is no folder .txt so some errors pop up.
Hi Authors, I have questions. during pip install -r requirements.txt, I'm unable to install the requirements file as it not available even I'm try to locate it. No luck. I run on Windows 11, the installation is similar to yours. Hope hearing news from you soon
same, did you find a solution?
same to did u find a solution? @@javierjp8549
@@javierjp8549 The file simply isn't there, can't find it in any of the versions, hence this "how to" is invalid
Anyone have any suggestions for fixing the error I'm getting when running "pip install -r requirements.txt" (note: I've tried running "pip3 install -r requirements.txt" and get the same error)
ERROR: Failed building wheel for llama-cpp-python
I am stuck on this one as well, but I know its related for not having a c++ compiler on windows.
Make sure you change on your ".env" file the following line
MODEL_TYPE=LlamaCpp
Also try to run everthing from a virtual environment, as this will reduce the chances of error.
Make sure you are logged in as Admin, or when you run command it will not write to files on Windows root directories
PrivateGPT only runs on Python version 3.10 or more.
Make sure you are using python3 when building the application
py -3 -m pip install -r .
equirements.txt
If you discover how to fix the issue please post here :)
@@cybersamurai99 Stuck at the same step, but I get the error, that pip is not a Cmdlet, a function, or script file.... Any thoughts on that? Have Python 2023.6.1 installed. Edit: Installed Python 3.11.4 from the website, doesn´t seem to change the error. Installed it in Visual Studio before (version 2023), but that seems not to be enough. Am I missing something? (On Win 10, PC) Edit2: Ok, after downloading Python from the Microsoft "Appstore", it finally worked.
Ok, now I made the install run, but got the same ERROR: Could not build wheels for llama-cpp-python, hnswlib, which is required to install pyproject.toml-based projects
Is it needed, or is it all ok, because its using GPT4ALL, not LlamaCpp ?
you have to install clang and after that it will work i had the same issue and i fixed it with that also do a lot of reboots after installations ...
@@ΚΩΝΣΤΑΝΤΙΝΟΣΠΑΠΗ Thanks a lot! Will try that out.
This is more complicated than that. You would need C++ complier, python, Visual Studio etc etc.
SO what is the secret to build it without getting any errors, because I am constantly getting llama-cpp-python error.
Does anyone know how to solve this on either linux or window?
Edit: the secret is to use venv or coda to build your code 👌
I ran chatdocs which is based on privategpt in a docker container. seems to run well too.
Sounds like a c++ compiler issue
Does anyone know if this works only with English language?
I'm not even sure yet he understands English. I am only getting stuff like this back "I don't have enough information to answer that question as it's not clear what the context is referring to or if there are any specific details provided in relation to a topic." 😂
is it possible to "train" the private gpt by connection to own datasource/database of existing app, or document in own NAS ?
how?
Nice info!
this doesnt work. after installing multiple missing libraries it gets stuck at ValueError: You are using a deprecated configuration of Chroma.
Hi, i installed everything and the code runs..but after entering the query, im not getting any response..whys that? I waited for hours..
Me, too. Does anyone have a solution for this? I'm using Windows and VS.
i thought git comes with your visual studio installation instruction
funny how it all seems so easy. In reality, this is a nightmare.
@Howfinity I receive an error saying that "No sentence-transformers model found with name sentence-transformers/all-MiniLM-L12-v2. Creating a new one with MEAN pooling". It then times out because I need to run this offline. Any answers?
Why do you need VS? I am not seeing a reason for that route.
It's VSCode and you don't. Can be done all at command line as long as you have Python and Git installed.
Could not open requirements file: [Errno 2] No such file or directory: 'requirements.txt'
can you use wsl (windows subsystem for linux) for windows?
*Some extra steps I took:*
1. Install Visual Studio. Not the Code. Just the Visual studio. Then install C++ development environment from there.
2. When ingesting, run
python ingest(dot)py
This guy wrote *python3* instead of *python*, which didn't work for me. python3 might be the code for your environment. But for me, that was just python
Note; add an actual fullstop where I said (dot) .
the dude used a Mac. Windows and Linux are different.
@@morris5648 ohh
Can you guide me change llamacpp parameters to reduce time of query
If you get this error when using this command poetry install --with ui,local
Group(s) not found: local (via --with), ui (via --with)
Use this command
poetry install --extras "ui vector-stores-qdrant llms-ollama embeddings-huggingface"
You're Welcome :)
There are many many steps missing on this instruction. Packages are missing, as is the requirements.txt from the repository. This video need update.
Only works in English? I mean the files must be on English? The questions also should be formulated on English ?
I give up. I have yet to find any step-by-step tutorial on anything involving python that works on my Mac. It's absolutely impossible. This time I got a full 4 minutes and 20 seconds in before any error messages which I think is a new record for me. 😕
Great Guidance! Thank you! Btw: I got this message ==> UnicodeDecodeError: 'charmap' codec can't decode byte 0x9f in position 15350: character maps to
Any help, please?
Great for people who know anything about CMD and how to use it, not so good for the regular users who need an EXE installer and a GUI.
why is my prompt different from yours?
How to add our data to private gpt?
Does it only work on Windows 10?
I installed Visual Studio Code, and it's not recognizing the 'git' command
[EDIT] Well, that was nothing, quick fix. However, can't install all of the dependencies, because of some C++ compiler error. Tried 2 solutions already, and not going forward...
Install C++ make sure it is configured so it set up with python. If not, uninstall python, and make sure you have the most recent version of both programs.
are you sure git is installed?
it's not working for me :( i'm stack on the "enter a query" step, i never receive an answer :(
Do you also get an "Index not found"-error? Did you resolve it, in the meantime?
It works, but it is slow and the results are not good enough. Nice to try for understanding and learn but not useful to work with from my humble point of view.
isn it depend on the PC specs ?
3 minutes in, we find out this is for installing on a MAC 😞
Mac. Yup
Mine looks nothing like that. Windows.
Is it possible to install more than one llm and use all of them at once?
Yes, but you need additional software, unless you want to go thought the tedious process of having to verify everything by hand and key, just to get 10 step in. However I am no expert.
Visual Studios layout is completly different for me
This is "Visual Studio Code" not "Visual Studio" which may be your issue. He did say the latter by mistake.
not workin, man!!!!!
I've watched tons of videos and do not understand.
We need an AI which finds tutorials at our level and on our OS lol
I'm half way through this, and you are still talking about requirements and where to get python. I think your course would be better served to assume that anyone tackling privategpt would already know about git, python, requirements file, and their environment. You could make separate videos on those topics and reference them. But so far I haven't learned anything about privategpt. Just meant to be constructive criticism.
But it’s only a 10 minute video and my entire channel is design for beginners. After making a 1000 videos, I can’t ever made assumption of previous knowledge. I’ve tried and it helps a lot less people.
@@Howfinity I Understand. I was making the suggestion though that you can do a separate video on the installation stuff that is generic and then focus your video on the actual topic. You can then just suggest they watch the setup video and then move on to your topic. I'm sure you will make more videos. Appreciate it!
Not everyone trying this works with python and Visual Studios on a daily basis. I'm familiar with them, from years ago. I don't have them install in my computer, and this has been the easiest to follow tutorial.
looks easy until you start installing
FJB
This is outdated video.
This tutorial is not only outdated but fraught with errors. The links have expired, the commands have changed, and the recent updates make following this frustrating and irrelevant. AVOID
Wwhhhaaaaa ??
This just seems lazy...ChatGPT I get because it does way more....if its your private documents you should know what u wrote and where it's stored....outside of the legal profession or industries that I'm ignorant about..just seems like one them things that's not gonna get used much but they throw in the bundle as a selling point...like the random stuff on Adobe..you only really need 2-3 things in it
No because some people need to use it to search through their documents during exam and exams have no Internet access
I think this is not a search tool to make search within your documents. online chatGPT is limited with tokens you can enter into chat window and what if you want to analyze some 500 pages long document and play with it?
can anyone help me
whenever i put this command pip install -r requirements.txt after some times i got an errro od meta failed
any instructions on how to get this working in a BSD jail AKA TrueNAS 12.3.4 @Howfinity
Anyone can help please.!! I am not able to install it on local PC. (missing requiremetns.txt and examples.env)
privateGPT> pip install -r requirements.txt
ERROR: Could not open requirements file: [Errno 2] No such file or directory: 'requirements.txt'
it looks like the 'requirements.txt' is not there.
Any thoughts on how to get past this?????