Пікірлер
@stephenzzz
@stephenzzz 6 күн бұрын
Thanks DevBytes
@scscyou
@scscyou 6 күн бұрын
Do you have any recommended setup for a computer where you don't want to risk running an untrusted binary? For example, unlike Android, on Windows, any application you run even once can have permanently taken control over your system, forever stealing all your accounts in all apps. One of the obvious solutions would be to only use web versions of the apps via browser, and to run both the web apps and your AI model servers in a Docker. If you have multiple computers, it would also be preferable to share their resources, which you'd probably control via Kubernetes to be able to describe and orchestrate a more complex setup... I'm feeling that tutorials are very lacking in this common sense way of requiring security. (Also, when running ollama in WSL2, my WSL stopped working with an error regarding insufficient resources, which blocked me from using it at all, so it's quite non-trivial to set up even a simple thing such as containerization nowadays with how everything works as a black-box.)
@AIDevBytes
@AIDevBytes 6 күн бұрын
There is a web version that you can use if you don't want to install it locally for "security" reasons. I'm running this on Mac so all binaries that are unsigned will be flagged by the system. Macs by default blocks all untrusted binaries and you must go into security and privacy to unblock the binary to run it. Not recommended unless you are the developer. Chatbox is a signed binary by a known developer so that is how it's available in the app stores. Also, any application trying to gain root access must request the access. I can't speak for Windows since I only use Mac or Linux and haven't focused on Windows in a long time.
@user-wr4yl7tx3w
@user-wr4yl7tx3w 6 күн бұрын
Is it free?
@AIDevBytes
@AIDevBytes 6 күн бұрын
👍 Correct, Chatbox is free software. But do offer paid offerings.
@yacu007
@yacu007 9 күн бұрын
amazing
@emilianofernandezcervantes4042
@emilianofernandezcervantes4042 10 күн бұрын
I spent around 4 hours trying to do this from Windows Subsystem for Linux. I'm glad to see that it is so much easier from MacOS.
@pinkhilichurl7670
@pinkhilichurl7670 12 күн бұрын
transferring model data Error: unsupported content type: text/plain; charset=utf-8 FROM llama3:8b PARAMETER temperature 1 PARAMETER stop <|start_header_id|> PARAMETER stop <|end_header_id|> PARAMETER stop <|eot_id|> PARAMETER stop <|reserved_special_token TEMPLATE """ {{ if .System }}<|start_header_id|>system<|end_header_id|> {{ .System }}<|eot_id|>{{ end }}{{ if .Prompt }}<|start_header_id|>user<|end_header_id|> {{ .Prompt }}<|eot_id|>{{ end }}<|start_header_id|>assistant<|end_header_id|> {{ .Response }}<|eot_id|> """ SYSTEM You are a helpful AI assistant named LLama3. Please provide the aspects discussed in the following Bengali product review. Please only consider the aspects from the following list of aspects: product, service, delivery, packaging, seller, rider, price, shelf. can you tell me why i am facing this error? I followed all the steps as you did.
@AIDevBytes
@AIDevBytes 12 күн бұрын
I haven't seen that error before. If I had to guess it would be because you have a multi-line SYSTEM message but it's not wrapped in triple quotes. Example: SYSTEM """Line one Line two Line three '''""
@ascaridesign
@ascaridesign 13 күн бұрын
Hello, it's works but only if I use terminal, with OllamaWebUi not, you know why ? Thx
@AIDevBytes
@AIDevBytes 13 күн бұрын
I have tried with OpenWeb UI and ran into the same issues. I'm assuming there is a bug in the OpenWebUI app that is causing the errors.
@dhaneshdutta
@dhaneshdutta 6 күн бұрын
@@AIDevBytes runs perfectly on mine..with docker
@toddroloff93
@toddroloff93 14 күн бұрын
Great Video. Thanks for doing this.
@taylorerwin807
@taylorerwin807 14 күн бұрын
For those having issues at around the 35min mark, you need to ensure that your role and function can access the internet aka. 'VPC'. Change those setting in the IAM menu for your user. Once that permission is added (AWSLambdaVPCAccessExecutionRole), you go back to configuration in the function and go down to VPC.
@aliceiqw
@aliceiqw 15 күн бұрын
Please help with this: **(venv) (base) iamal@IAMALs-MBP suede % ollama create my-llama3-model -f /Users/iamal/Desktop/suede/custom-llama3.yaml** Error: command must be one of "from", "license", "template", "system", "adapter", "parameter", or "message"
@AIDevBytes
@AIDevBytes 15 күн бұрын
What does your file look like? Also, on the model files I created I do not use an extension.
@aliceiqw
@aliceiqw 15 күн бұрын
when running this in the terminal: "ollama create my-llama3-model -f custom-llama3.yml" I get this error: Error: command must be one of "from", "license", "template", "system", "adapter", "parameter", or "message"
@AIDevBytes
@AIDevBytes 13 күн бұрын
What does your model file content look like?
@kannansingaravelu
@kannansingaravelu 16 күн бұрын
How do we fine tune the model? Do you have a video on it?
@AIDevBytes
@AIDevBytes 12 күн бұрын
I currently don't have a video covering that yet, but looking to create one in the future.
@tnsgaming6571
@tnsgaming6571 17 күн бұрын
Which extension do you use to create the custom -llama3 Pls help
@AIDevBytes
@AIDevBytes 15 күн бұрын
I don't add an extension to the model file when I create it.
@RNMSC
@RNMSC 17 күн бұрын
It's not universal across Linux, but many distributions have depricated the ifconfig command for getting the network ip address. One command you can use to get this is 'ip address show | grep inet' which should give you a list similar to ifconfig on the mac, and will include ipv6 addresses if you have ipv6 enabled. Lan specific ipv6 addresses will begin 'fe80:' and should be available to all hosts within the local network. If you have ipv6 addresses other than this and '::1' you may want to make sure that your network is blocking external traffic for any services you don't want to share outside of your lan. At the moment, my openUI service does not appear to be attaching to the IPv6 stack, however that is potentially subject to change in the future. You can always ask an llm how to connect to a server's ipv6 address and you should get some possibly useful responses, however if the service isn't attaching to ipv6, it's not going to work. Now if I could find a way to get OpenUI on a docker server, at a locally significant hostname, to connect to a system running ollama (because while the docker server doesn't have a graphics card capable of givving happy response rates, the game system does, and can be used while I'm not gaming for that. However it looks like ollama isn't attaching to any of the hardware nic ipv4 addresses so far. I'll get it worked out eventually. (maybe?)
@RNMSC
@RNMSC 17 күн бұрын
Solved that. There is an 'error message' in the output after editing the systemctl file for the ollama.service that suggests that it didn't take. I forced a copy of the teporary output file to the output file, then followed up with the systemctl daemon-reload, and systemctl restart ollama.service commands (sudo'ed) and it appears to have taken. Now to get openui to point at that. Adding the server to connections under setup seems to have no effect. Ah well. may have to remove and re-install the docker client on my localhost.
@davidtindell950
@davidtindell950 20 күн бұрын
thank u. new subscriber !
@stanleykwong6438
@stanleykwong6438 22 күн бұрын
Thanks for acting on my suggestion and making this video. As always, great content. Keep up the great work!
@AIDevBytes
@AIDevBytes 22 күн бұрын
thank you for the suggestion. I always read the comments to see what viewers such as yourself would like to see. Hope you found the video helpful.
@stanleykwong6438
@stanleykwong6438 21 күн бұрын
@@AIDevBytes Very helpful, i was able to setup the hosting for my wife to use on her computer with this. Side note for future topics when you have time and it aligns to your interest. 1. Step by step incorporating stable diffusion into OpenWebUI would be great 2. How to do RAG properly (from the best system in your opinion, to best practices on how to structure your format for better ingestion) 3. How to allow external access securely? I've seen some videos where Ngrok is used, but their walkthrough was confusing. Thanks again.
@AIDevBytes
@AIDevBytes 21 күн бұрын
Thank for the suggestions
@AIDevBytes
@AIDevBytes 22 күн бұрын
Thank you @stanleykwong6438 for the video idea. Here are other Open WebUI videos you may like: ULTIMATE Llama 3 UI: Dive into Open WebUI & Ollama! kzbin.info/www/bejne/emWrZpuDgtSNaJo ULTIMATE Llama 3 UI: Chat with Docs | Open WebUI & Ollama! (Part 2) kzbin.info/www/bejne/oXXadnydotaUe6c Open WebUI & OpenAI DALL-E 3: Effortless Text to Image Generation kzbin.info/www/bejne/enm5gmSLd8aMg7M
@engdoretto
@engdoretto 22 күн бұрын
Thanks for the informative video. Would it be possible for me to chat with my own pdf documents?
@AIDevBytes
@AIDevBytes 22 күн бұрын
Glad you found the video helpful. Yes, check out this video here kzbin.info/www/bejne/oXXadnydotaUe6c
@Panda-sv3oy
@Panda-sv3oy 25 күн бұрын
My friend, I love your videos and I wanted you to explain to us how to use n8n .. I want to use it to automate everything possible in my daily use
@yurijmikhassiak7342
@yurijmikhassiak7342 26 күн бұрын
Why you are showing us UI with completely oh "hellow word" code. Is it possible to do something harder? Can you change the design with feedback? Can you setup the style library to have consistent components?, generate a few options to select from?
@AIDevBytes
@AIDevBytes 26 күн бұрын
This video is meant to just to show how to get OpenUI up and running. I didn't want to start with anything too complex, since many people who watch the channel are new to many of these tools. I usually create more advanced follow up videos later. Yes, you can change the style with feedback by continuing with a follow on prompt and the output will be altered There is not presently a settings to keep the style consistent. I would personally only use this app for prototyping and then further refine the design in something like VSCode. Also, the model you use is going to dictate the quality of output. I find that using the OpenAI GPT API generates pretty good UI code, but I like to focus on free things on the channel. I may create a video later using Groq to generate the UI.
@lethalviper250
@lethalviper250 26 күн бұрын
This works for HTML only or can we use it to prepare Java and Python based UI?
@AIDevBytes
@AIDevBytes 26 күн бұрын
You could try to change the system prompt to see if it will create python or Java code for you, but it will not render a preview for you. Since Java and Python UIs aren't HTML based natively. This tool is meant to primarily be used developers that by develop who use web frameworks that use HTML and JSX as their native UI output.
@tatkasmolko
@tatkasmolko 27 күн бұрын
And does generating html component from image via ollama work for you? I get an Error! 500 Internal Server Error: object of type bytes is not JSON serializable
@AIDevBytes
@AIDevBytes 27 күн бұрын
You will need to download the Llava model into Ollama. Then set that as your model in OpenUI. I will say that using an image to generate the UI does not give great results based on my testing.
@tatkasmolko
@tatkasmolko 26 күн бұрын
@@AIDevBytes Thank you for your help. I didn't read the documentation very carefully :D I was using the wrong version of the llava model. Everything works. Question is there any model from open source portfolio that can render html from image in GPT-3.5-Turbo quality? For example some model via LM Studio?
@AIDevBytes
@AIDevBytes 26 күн бұрын
Glad you go it working. I had sent another message but deleted it since I realized that would not work because we need a better open-source multi-modal first to generate better HTML
@tatkasmolko
@tatkasmolko 26 күн бұрын
@@AIDevBytes I skipped docker and used directly the frontend application vitejs with a little code modification I was able to connect it to LM Studio (windows) where there is a greater possibility of open source models with Vision Adapter and I achieve much better results than through docker and ollama llava model. The advantage is that with LM Studio I can use one API called Multi Model Session which allows me to achieve better results when generating html code from images without the need to use the paid version of GPT-4o. But need high VRAM of GPU.
@AIDevBytes
@AIDevBytes 26 күн бұрын
Thanks for the info. I will have to try it out.
@stephenzzz
@stephenzzz 27 күн бұрын
Thanks DevBytes
@AIDevBytes
@AIDevBytes 28 күн бұрын
🧑‍💻 Commands for setting up OpenUI Be sure you have Python installed: www.python.org/downloads 🧑‍💻 Ollama Mac Setup: kzbin.info/www/bejne/ZmStkI1se7-LipY 🪟 Ollama Windows Setup: kzbin.info/www/bejne/e2bOkHyio9KYnsU OpenUI Setup Commands 👇 Command 1: git clone github.com/wandb/openui Command 2: cd openui/backend Command 3: pip install . Command 4: export OPENAI_API_KEY=xxx Command 5: python -m openui
@stanleykwong6438
@stanleykwong6438 Ай бұрын
i've been trying to get my own models setup and your series of videos have been exceptionally clear, please post more videos for an indepth use of Open WebUI and perhaps for the following so that non-technical people like me can implement it. 1. How to setup and allow other computers with WebUI installed to access the computer that has ollama installed on the same network. 2. How to setup "My Modelfiles" within WebUI, specifically how the Content section of the Modelfile is compiled, and how to incorporate other parameters beyond temperature. Exceptional work! I will spread the word with my team. Keep it up.
@AIDevBytes
@AIDevBytes 29 күн бұрын
I will look creating separate videos on those two topics. Thanks for the feedback.
@primordialcreator848
@primordialcreator848 Ай бұрын
any chance you can please make one using Automatic1111 connecting to a local instance of stable diffusion or any open source image generator please? I've been trying to get it working but seems to get stuck on saving when ever I try to connect to the local instance with api mode active. any info on this would be amazing! thank you for another great video! <3 <3 <3
@AIDevBytes
@AIDevBytes Ай бұрын
Glad you found the video helpful. I will have to look at getting stable diffusion working correctly on my M3 Mac. It was not the best experience the last time I ran it.
@primordialcreator848
@primordialcreator848 Ай бұрын
@@AIDevBytes thanks for the reply! - got stable diffusion 2.1 working on my 1max mac, I'm able to generate images with it by it self but stuck having it integrate with WebUI. i can use Dall-E 3 for now np. just hopping for something free to use lol
@ph3ll3r
@ph3ll3r Ай бұрын
Where can get the ollama run ollama: instruct model? The closest that I found was ollama run meta-llama/Meta-Llama-3-8B-Instruct which does not work form hugging faces
@AIDevBytes
@AIDevBytes Ай бұрын
Check out these vdeos here: Windows Setup - OLLAMA | Want To Run UNCENSORED AI Models on Windows?: kzbin.info/www/bejne/e2bOkHyio9KYnsU Mac Setup: OLLAMA | Want To Run UNCENSORED AI Models on Mac (M1/M2/M3): kzbin.info/www/bejne/ZmStkI1se7-LipY Models can be found here ollama.com/library
@eviltoast3r
@eviltoast3r Ай бұрын
Ollama run llama3:instruct will pull and run the latest instruct model
@AIDevBytes
@AIDevBytes Ай бұрын
Correct, if you just want to pull the model run -> ollama pull llama:instruct
@VjRocker007
@VjRocker007 Ай бұрын
I got it set up! Thank you for the quick guide!! Dev would you make a video on how to setup lobechat as well? It's a very interesting piece of front end. But hosting locally, I found it quite challenging since I'm not too technical in self hosting apps on the Mac.
@AIDevBytes
@AIDevBytes Ай бұрын
I'll take a look at lobechat. I have not personally used it, but it sounds interesting.
@madhudson1
@madhudson1 Ай бұрын
How much on average does it cost to generate. I know itll be based on tokens. Just after a ballpark
@AIDevBytes
@AIDevBytes Ай бұрын
I have updated the description with Chapters. Jump to the chapter "DALLE-3 Pricing" in the video. I explain pricing there. In the end your cost is going to be determined by the usage of the DALLE-3 API. The more the use the API the more it will cost. Also, note image generation is not tied to tokens like text generation. I am using free models for text generation in this video using Ollama and the Llama model. Videos for setting up Ollama can be found here: Windows Setup kzbin.info/www/bejne/e2bOkHyio9KYnsU Mac Setup: kzbin.info/www/bejne/ZmStkI1se7-LipY
@tiffanyw3794
@tiffanyw3794 Ай бұрын
Can you show how to connect sdforge it’s for computers with lower graphics
@AIDevBytes
@AIDevBytes Ай бұрын
I'll have to take a look at it. I know people seem to use it with Stable Diffusion. I personally don't do a lot with Stable Diffusion because of the hardware requirements that most people need and the clunkiness of the UI.
@tiffanyw3794
@tiffanyw3794 Ай бұрын
@@AIDevBytes sdforge was created by the a1111 it works great for low gpu users would be awesome to use in this Ui. I see it only has 3 options. Is there away to add a different image model
@AIDevBytes
@AIDevBytes Ай бұрын
Since OpenWeb UI is an open-source project anyone can edit the code. But there is no way to currently add model to the image section. You can add new model for chat. If you check out video 1 in the series I setup OpenWeb UI with Ollama kzbin.info/www/bejne/emWrZpuDgtSNaJosi=QFr1LzQhONF4mbgp
@AIDevBytes
@AIDevBytes Ай бұрын
Thank you @vjrocker007 for the video idea. Part 1 & 2 of this series can be found below. 📺 PART 1 - ULTIMATE Llama 3 UI: Dive into Open WebUI & Ollama! kzbin.info/www/bejne/emWrZpuDgtSNaJo 📺 PART 2 - ULTIMATE Llama 3 UI: Chat with Docs | Open WebUI & Ollama! kzbin.info/www/bejne/oXXadnydotaUe6c
@stephenzzz
@stephenzzz Ай бұрын
Thanks!
@AIDevBytes
@AIDevBytes Ай бұрын
You bet!
@LesCalvin3
@LesCalvin3 Ай бұрын
I like the clear presentation a lot. Thanks.
@MaliciousCode-gw5tq
@MaliciousCode-gw5tq Ай бұрын
Hi can you provide the specs of your GPU use for this project to get better experience and smooth response.
@AIDevBytes
@AIDevBytes Ай бұрын
Check out the video description. I have my Mac book specs below.
@MaliciousCode-gw5tq
@MaliciousCode-gw5tq Ай бұрын
@@AIDevBytes Awesome thanks. It will help get idea for a RIG Thanks Man Keep it up!...
@alphaobeisance3594
@alphaobeisance3594 Ай бұрын
I can't seem to comprehend how to provide access to my AI through OpenWebUI via my domain. I'd like to grant access for my family but I for the life of me can't seem to get it set up for public access.
@AIDevBytes
@AIDevBytes Ай бұрын
In this video, I demonstrate local access only. I will explore how to host this on a server and create a follow-up video on the process. Note that hosting on a server will incur costs due to GPU pricing.
@alphaobeisance3594
@alphaobeisance3594 Ай бұрын
@@AIDevBytes I've got homelab and hardware. Just can't seem to figure out the networking side of things for some reason. Tried proxy through Apache but it must be over my head as I can't get it to function correctly.
@AIDevBytes
@AIDevBytes Ай бұрын
Gotcha, yeah unfortunately I wouldn't really be able to help there since I don't know your network configuration or topology. You should be able to go into your router settings and setup port forwarding to you homlab server.
@Thedemiul
@Thedemiul Ай бұрын
Hi, I noticed your job post on Upwork about hiring an editor, and I'm interested in that position. Could we schedule a chat to discuss further?
@AIDevBytes
@AIDevBytes Ай бұрын
Please send all proposals through Up work and someone on the team will reply there. Thanks
@Thedemiul
@Thedemiul Ай бұрын
@@AIDevBytesBoss, I've run out of connects there.
@AIDevBytes
@AIDevBytes Ай бұрын
Sorry the individual that handles openings only accepts proposals through that platform unfortunately.
@AIDevBytes
@AIDevBytes Ай бұрын
Feel free to send your Upwork username and we can send you a proposal request. You can email here [email protected] Thanks
@VjRocker007
@VjRocker007 Ай бұрын
Amazing video! thank you for sharing! Bytes could do a video tutorial on setting up image generation from Dall-E through webUI as well? That would be really helpful.
@AIDevBytes
@AIDevBytes Ай бұрын
Glad you liked the video. I'll look into creating a video based on your suggestion.
@VjRocker007
@VjRocker007 Ай бұрын
You're a legend, thank you for posting these how to videos. Keep it up, for a person who doesn't have time to go through all of these docs these have been incredibly helpful. Appreciate all of your hard work!!
@hamzahassan6726
@hamzahassan6726 Ай бұрын
hi, I am trying to make a model file with these configurations: # Set the base model FROM llama3:latest # Set custom parameter values PARAMETER num_gpu 1 PARAMETER num_thread 6 PARAMETER num_keep 24 PARAMETER stop <|start_header_id|> PARAMETER stop <|end_header_id|> PARAMETER stop <|eot_id|> # Set the model template TEMPLATE "{{ if .System }}<|start_header_id|>system<|end_header_id|> {{ .System }}<|eot_id|>{{ end }}{{ if .Prompt }}<|start_header_id|>user<|end_header_id|> {{ .Prompt }}<|eot_id|>{{ end }}<|start_header_id|>assistant<|end_header_id|> getting Error: unexpected EOF Could you tell me what am I doing wrong?
@AIDevBytes
@AIDevBytes Ай бұрын
Looks like you didn't close your double quotes at the end of your template. Simple mistake which can drive you crazy 😁 Let me know if that fixes your issue. EDIT: Also, use triple quotes like this when using multiple lines for your template. TEMPLATE """ Template values goes here """
@hamzahassan6726
@hamzahassan6726 Ай бұрын
@@AIDevBytes getting same error with this # Set the base model FROM llama3:latest # Set custom parameter values PARAMETER num_gpu 1 PARAMETER num_thread 6 PARAMETER num_keep 24 PARAMETER stop <|start_header_id|> PARAMETER stop <|end_header_id|> PARAMETER stop <|eot_id|> # Set the model template TEMPLATE """ {{ if .System }}<|start_header_id|>system<|end_header_id|> {{ .System }}<|eot_id|>{{ end }}{{ if .Prompt }}<|start_header_id|>user<|end_header_id|> {{ .Prompt }}<|eot_id|>{{ end }}<|start_header_id|>assistant<|end_header_id|> """
@AIDevBytes
@AIDevBytes Ай бұрын
When I get some free and I'm at my computer again today. I will give it a try to see if I can isolate the problem and let you know.
@hamzahassan6726
@hamzahassan6726 Ай бұрын
@@AIDevBytes thanks mate. much appreciated
@AIDevBytes
@AIDevBytes Ай бұрын
@@hamzahassan6726 I copied the model file content you had and pasted into a new file and was able to create a new model. I am not quite sure why you are the getting the error: "Error: unexpected EOF". I have not been able to duplicate the error. Also, one thing to call out looks like you are not using the llama3 template from ollama, but that doesn't appear to be causing the issue. I would make sure you are not using rich text format in your model file and ensure that it is plaintext only. if you go to the llama3 model (ollama.com/library/llama3:latest/blobs/8ab4849b038c) the template looks like this: {{ if .System }}<|start_header_id|>system<|end_header_id|> {{ .System }}<|eot_id|>{{ end }}{{ if .Prompt }}<|start_header_id|>user<|end_header_id|> {{ .Prompt }}<|eot_id|>{{ end }}<|start_header_id|>assistant<|end_header_id|> {{ .Response }}<|eot_id|>
@elrilesy
@elrilesy Ай бұрын
Hey mate, this is absolutely awesome, thanks so much. I'm wondering how well this would work at Knowledge Bases? Ie lets say I'm a decent sized company that has a huge internal wiki covering all sorts of internal processes that guides our employees in how to operate the business. Do you think I could upload all our wiki articles to the documents section and then give this to our team to be able to talk to our own internal chatbot to ask questions about what they should do in certain situations? Or do you think uploading very large amounts of documents would sort of overwhelm it and not produce great results? i.e do you think to achieve this outcome you'd actually need to fully retrain the base llama3 model itself rather than just adding documents on top of an existing stock llama3 model? Because as far as I can tell in my laymands understanding of all this is that when you upload documents, each prompt you enter it's having to kinda look through it all for an answer where as if you re-trained the base model it would know it more "naturally" haha. Sorry, not sure if I've explained that very well.
@AIDevBytes
@AIDevBytes Ай бұрын
for this architecture you would not need to to retrain the model, since it's using RAG (Retrieval Augmented Generation) to find the answers in the documents. I will say this solution is really for users to run it locally on their computer, since the document store is running in a container on the users computer. This solution would not probably scale well for multiple users across multiple computers. You would want a more enterprise or custom solution if you are going to run this for your business with multiple users on multiple computers.
@FaeEvergreen
@FaeEvergreen Ай бұрын
I tried this setup but the RAG in WebUI is absolutely atrocious. Maybe I'm just not having a great time because I'm on a Windoze machine, or maybe I messed up somewhere but my conversations seem to have VERY hit & miss accuracy when working with docs. I have tried compiling all docs into a master doc, I've tried using multiple small docs. I've tried .doc, .md, .txt, .csv with varying levels of success. It just seems like it's great in theory and not so much in practice. Like it's really cool that it's available and working, but as far as using it in my every day life -- probably not.
@ssswayzzz
@ssswayzzz Ай бұрын
do you think you will need more than m3 max 36gb 14cpu/30gpu for running all these models im thinking of getting one myself to try all these models all i know that i need a capable device , btw thank you so much for your videos once i get this device im sure i will be around your channel alot benefiting from your experience thank you
@AIDevBytes
@AIDevBytes Ай бұрын
I don't have to have the M3 Max, but it definitely will help. I say the minimum RAM you will need on Apple Silicon is 16 GB to run the smaller models. With the M3 Max you will be able to run most of the Open-Source models up to about 40 billion parameters, just not the really large parameter models like the 70 billion parameter models. Also, happy you are finding the content useful.
@art3mis635
@art3mis635 Ай бұрын
when running the docker command I get this error: %2Fdocker_engine/_ping": open //./pipe/docker_engine: The system cannot find the file specified. See 'docker run --help'. any help ?
@AIDevBytes
@AIDevBytes Ай бұрын
Did you already have docker installed or did you install a new version of docker?
@art3mis635
@art3mis635 Ай бұрын
@@AIDevBytes yes I have docker desktop installed, but there is a newer version. Should I update ? Current version: 4.29.0 (145265) New version: 4.30.0 (149282)
@AIDevBytes
@AIDevBytes Ай бұрын
I recommend running the latest version to see if that fixes the issue. I am running the newest version of Docker.
@art3mis635
@art3mis635 Ай бұрын
@@AIDevBytes it worked thank you
@AIDevBytes
@AIDevBytes Ай бұрын
glad that fixed your issue
@eevvxx80
@eevvxx80 Ай бұрын
Thanks mate, I have a question. Can I add my text to llama3?
@AIDevBytes
@AIDevBytes Ай бұрын
Can you explain further? Do you mean add you own text to the SYSTEM parameter? Not sure I'm am following your question.
@stephenzzz
@stephenzzz Ай бұрын
Brilliant thanks!
@HernanMartinez82
@HernanMartinez82 Ай бұрын
Which embedding model did you use for the document chat?
@AIDevBytes
@AIDevBytes Ай бұрын
By default the HuggingFace sentence transformer model is used (sentence-transformers/all-MiniLM-L6-v2) for embedding. You can change it to use whatever model you would like by going to your "Document Settings".
@victor_ndambani
@victor_ndambani Ай бұрын
thank u man you now have a new sub
@AIDevBytes
@AIDevBytes Ай бұрын
PART 1 can be found here -> ULTIMATE Llama 3 UI: Dive into Open WebUI & Ollama! kzbin.info/www/bejne/emWrZpuDgtSNaJo
@laalbujhakkar
@laalbujhakkar Ай бұрын
So, what's the point of "customizing" when I can just change the system prompt? Isn't it like copying /bin/ls to /bin/myls and feeling like I accomplished something?
@AIDevBytes
@AIDevBytes Ай бұрын
This a very simple example, but the purpose would be if you wanted to change multiple parameters as part of the model and use it in another application. Example, you could use the model with something like Open WebUI and then lock users into only using the model you customized with your new parameters.
@conerwei6720
@conerwei6720 Ай бұрын
my computer did now have good gpu,can you make a video about how wo use cloud ollama and local webui?
@conerwei6720
@conerwei6720 Ай бұрын
did not have good gpu,sorry
@AIDevBytes
@AIDevBytes Ай бұрын
Yes, I'll create a video of how to run Ollama in a cloud environment. I'll see how to setup Open WebUI also in a cloud environment. Probably two different videos at later dates.