- Видео 56
- Просмотров 138 614
JarvisLabs AI
Индия
Добавлен 28 янв 2020
Jarvislabs.ai provides a 1-click GPU cloud platform for your AI and DL training.
Spin up modern Nvidia A100, A6000, A5000, Quadro RTX 5000, Quadro RTX 6000 GPU, and more at lower prices in seconds.
Some of the prominent features are
🚀 Get an optimized environment of the popular frameworks (PyTorch, Tensorflow, Fastai, or BYOC).
🚀 Access JupyterLab configured securely.
🚀 SSH/connect to the instance using your favorite IDE.
💰 Pay per minute.
🚀 Scale GPUs, storage, and change GPUs anytime you want.
⏸️ Automatically pause instance through code.
🚀 We offer lower pricing per hour for interruptible workloads and reduced weekly/monthly pricing options for long-term dedicated use.
💬 Talk anytime with the team who built the product. We would love to help you on your Deep learning journey.
🤝 Trusted by 10,000's AI practitioners worldwide.
Spin your DL instance today at cloud.jarvislabs.ai.
Learn more at jarvislabs.ai/docs.
Spin up modern Nvidia A100, A6000, A5000, Quadro RTX 5000, Quadro RTX 6000 GPU, and more at lower prices in seconds.
Some of the prominent features are
🚀 Get an optimized environment of the popular frameworks (PyTorch, Tensorflow, Fastai, or BYOC).
🚀 Access JupyterLab configured securely.
🚀 SSH/connect to the instance using your favorite IDE.
💰 Pay per minute.
🚀 Scale GPUs, storage, and change GPUs anytime you want.
⏸️ Automatically pause instance through code.
🚀 We offer lower pricing per hour for interruptible workloads and reduced weekly/monthly pricing options for long-term dedicated use.
💬 Talk anytime with the team who built the product. We would love to help you on your Deep learning journey.
🤝 Trusted by 10,000's AI practitioners worldwide.
Spin your DL instance today at cloud.jarvislabs.ai.
Learn more at jarvislabs.ai/docs.
Google NotebookLM Audio Overview: Can AI Create Engaging Podcasts?
Google's new Audio Overview feature in NotebookLM claims to transform how we consume content by turning any document into a podcast-style conversation. But does it live up to the hype? In this video, we'll test this AI-powered tool to find out!
Here's what we’ll cover:
● How to access and use Audio Overview
● Testing it with a PyTorch blog post
● The pros and cons of the feature
● Potential issues with AI-generated content, like hallucinations
Watch to see if Google's latest AI innovation is the future of content consumption or just another gimmick. Don't miss out!
#GoogleAI #AudioOverview #NotebookLM
Here's what we’ll cover:
● How to access and use Audio Overview
● Testing it with a PyTorch blog post
● The pros and cons of the feature
● Potential issues with AI-generated content, like hallucinations
Watch to see if Google's latest AI innovation is the future of content consumption or just another gimmick. Don't miss out!
#GoogleAI #AudioOverview #NotebookLM
Просмотров: 612
Видео
Accelerating LLMs 10x with Pure PyTorch: No Custom Libraries
Просмотров 1 тыс.2 месяца назад
Discover how to supercharge your Large Language Models with native PyTorch! In this video, we break down the techniques from the PyTorch team's recent blog post "Accelerating Generative AI with PyTorch II: GPT, Fast". 🚀 Learn about: ● Native PyTorch optimizations for a 10x speed increase ● Reduce CPU overhead with torch.compile ● Int8 weight-only quantization ● Speculative decoding ● Tensor Par...
Run Llama 3.1 70B on H100 using Ollama in 3 simple steps | Open WebUI
Просмотров 1 тыс.2 месяца назад
Learn how to run the powerful Llama 3.1 70B language model on H100 GPU using Ollama in just three easy steps. We'll also show you how to use the Open WebUI for a user-friendly interface to use your model. 👍 Like, subscribe, and hit the notification bell for more AI model insights. 💬 Share your thoughts and questions in the comments below. Check out our socials: 🌐 Website: jarvislabs.ai/ 🎭 Disco...
What is Instruction Finetuning? | EP-1 | JarvisLabs
Просмотров 3012 месяца назад
Hey there! Welcome to our "What Is" series, where we break down AI concepts without the fancy jargon. In this quick 5-minute video, we're tackling Instruction Finetuning in AI. Don't worry if that sounds complicated - we promise it won't be! Here's what you'll learn, jargon-free: 1. The difference between regular AI models and instruction-finetuned ones 2. The special ingredients (data) these m...
Launch Ollama Serverless on JarvisLabs: A Step-by-Step Guide
Просмотров 1782 месяца назад
Deploy Ollama serverless for efficient LLM tasks. We'll learn how to deploy, manage, Ollama serverless on JarvisLabs in this video. Check out Ollama serverless: jarvislabs.ai/serverless/frameworks/ollama_serverless 👍 Like, subscribe, and hit the notification bell for more AI model insights. 💬 Share your thoughts and questions in the comments below. Check out our socials: 🌐 Website: jarvislabs.a...
Cursor: Build an AI App on GPU | JarvisLabs AI
Просмотров 3762 месяца назад
In this video we will learn how to use Cursor to build an AI app using a simple prompt on JarvisLabs. Download Cursor: www.cursor.com/ Check out our socials: 🌐 Website: jarvislabs.ai/ 🎭 Discord: discord.gg/sps7qeDKJr
Microsoft Florence 2 - Is it the best open source foundational vision model?
Просмотров 2712 месяца назад
In this video we will explore Microsoft's latest open-source computer vision model, Florence 2. Is it the best foundational vision model available? Chapters 0:0 Introduction 1:40 Training Dataset 2:38 Demo 6:38 Examples 10:12 Data Annotation 12:12 Model Architecture 13:42 Key Capabilities 15:02 Outro In this video, we'll look at: 🔍 Key features and capabilities of Florence 2 📊 How it compares t...
Flux LoRA: Finetune with images
Просмотров 1,8 тыс.2 месяца назад
Easily LoRA finetune Flux Dev! In this beginner-friendly tutorial, you'll learn how to finetune the popular flux dev model using your own images. Chapters 0:00 Intro 0:40 Instance specs & ComfyUI 1:45 Flux Models 2:50 AI-toolkit 3:35 Finetune - setup 4:50 Flux Blogs 5:10 Flux origin & capabilities 7:13 Finetune - setup 10:52 Finetune 11:15 Easy Captioning 11:57 Logs 12:15 Trigger words 12:32 Ad...
Create Your AI Twin | Kohya SDXL LoRA Tutorial: A Beginner's Guide
Просмотров 3,1 тыс.4 месяца назад
In this easy-to-follow tutorial, you'll discover: How to finetune the SDXL base model with just 10-15 of your images Step-by-step guide to set up and train your custom AI portrait generator Key parameters to modify for best results Ideal for: Stable Diffusion newcomers, AI art enthusiasts, and anyone curious about creating personalized AI portraits. ⏱️ Process breakdown: Setup time: ~15 minutes...
ComfyUI - Using Whisper speech node to make images talk(lip sync)! | Tutorial
Просмотров 6 тыс.6 месяцев назад
Hello everyone, in this video we will learn how to make images talk. Yes, you read that right. We will see how to download and install @impactframes 's custom nodes Whisper speech node and Dreamtalk to animate images. We will also build the workflow from the scratch and animate a picture of Ryan Reynolds. Check out Impact Frames on github: github.com/if-ai Workflow : github.com/jarvislabsai/com...
How to run llama3 8B & 70B using ollama on JarvisLabs | A quick walkthrough
Просмотров 5007 месяцев назад
Hello everyone, llama3 8B and 70B are the talk of the town in the recent days. We at JarvisLabs have made loading them in ollama faster than before. This is a quick walkthrough demonstrating how you can inference llama3 in ollama on JarvisLabs. Run ollama on JarvisLabs : jarvislabs.ai/templates/ollama Check out our socials: Website: jarvislabs.ai/ Discord: discord.gg/sps7qeDKJr X: j...
ComfyUI: Master Face Swap & Pose Mimicry with IP-Adapter v2 & ControlNet | JarvisLabs
Просмотров 8 тыс.7 месяцев назад
Hello everyone, In this video we will learn how to use IP-Adapter v2 and ControlNet to swap faces and mimic poses in ComfyUI. We'll walk through the steps to set up these tools in ComfyUI and demonstrate their capabilities. We will also go through practical examples to learn what should be done and how mistakes can be avoided. Enjoy!! Workflow : github.com/jarvislabsai/comfyui_workflows/blob/ma...
ComfyUI - Upscale Images upto 8K with SUPIR & Restore Images | JarvisLabs
Просмотров 5 тыс.7 месяцев назад
Hello everyone! In this video, we're going to look at how to upscale images to 8K resolution using SUPIR. SUPIR is a tool that has become popular for upscaling images to very high resolutions. We'll go through the features of SUPIR and show you how to use it to increase your images to 8K. By the end, you'll know how to get much higher resolution and detail in your images with SUPIR. Workflow : ...
ComfyUI - Learn how to generate better images with Ollama | JarvisLabs
Просмотров 20 тыс.7 месяцев назад
In this video we will learn how to use the power of LLMs using Ollama and Comfy_IF_AI nodes to generate the best images. Vishnu will also take us through on how to set up ollama in JarvisLabs' instances. Workflow : github.com/jarvislabsai/comfyui_workflows/blob/main/ollama_comfyui.json Check out Ollama: ollama.com/ Check out our ComfyUI basics playlist: ruclips.net/video/GUm0JJ57KC8/видео.html&...
ComfyUI - Getting started (part - 4): IP-Adapter | JarvisLabs
Просмотров 8347 месяцев назад
Hello everyone, in our last episode we saw how to build a Stable Cascade workflow. In this video we will see how to use IP-Adapter to have images as prompts, extract style from one image to another and much more! Make sure to like and subscribe to our channel for more ComfyUI videos. Chapters: 0:00 Intro 0:33 Image2Image (first workflow) 2:30 Style Transfer (second workflow) 5:14 Face Swap v1(t...
ComfyUI - Getting Started (part-3): Stable Cascade | JarvisLabs
Просмотров 5017 месяцев назад
ComfyUI - Getting Started (part-3): Stable Cascade | JarvisLabs
ComfyUI - Getting started (part - 2) : SDXL refiner & ControlNet | JarvisLabs
Просмотров 1,1 тыс.7 месяцев назад
ComfyUI - Getting started (part - 2) : SDXL refiner & ControlNet | JarvisLabs
ComfyUI - Getting started: basics (part -1) | JarvisLabs
Просмотров 1,1 тыс.7 месяцев назад
ComfyUI - Getting started: basics (part -1) | JarvisLabs
How to Finetune LLM with Axolotl | Tutorial | Walkthrough on JarvisLabs.ai
Просмотров 2,8 тыс.8 месяцев назад
How to Finetune LLM with Axolotl | Tutorial | Walkthrough on JarvisLabs.ai
Can AI Bet Google? Perplexity vs Sundar Pichai: The AI Battle You Didn't Know About!
Просмотров 2 тыс.10 месяцев назад
Can AI Bet Google? Perplexity vs Sundar Pichai: The AI Battle You Didn't Know About!
The architecture of mixtral8x7b - What is MoE(Mixture of experts) ?
Просмотров 4,2 тыс.10 месяцев назад
The architecture of mixtral8x7b - What is MoE(Mixture of experts) ?
Apple Ferret a Multimodal LLM: The First Comprehensive Guide (Quick Demo with steps)
Просмотров 7 тыс.10 месяцев назад
Apple Ferret a Multimodal LLM: The First Comprehensive Guide (Quick Demo with steps)
Exploring the fastest open source LLM for inferencing and serving | VLLM
Просмотров 10 тыс.10 месяцев назад
Exploring the fastest open source LLM for inferencing and serving | VLLM
[Stable diffusion webUI]AUTOMATIC1111 : Learn Text to image under 10 minutes
Просмотров 50210 месяцев назад
[Stable diffusion webUI]AUTOMATIC1111 : Learn Text to image under 10 minutes
First Impressions: Our Unfiltered Take on Microsoft Copilot!
Просмотров 23810 месяцев назад
First Impressions: Our Unfiltered Take on Microsoft Copilot!
Invisible Art: How to Embed Texts and Logos in Images Using ComfyUI | JarvisLabs AI Tutorial
Просмотров 92211 месяцев назад
Invisible Art: How to Embed Texts and Logos in Images Using ComfyUI | JarvisLabs AI Tutorial
Understanding Fastai Vision Learner
Просмотров 1,6 тыс.2 года назад
Understanding Fastai Vision Learner
Understanding Fastai Learner Class Part - 2
Просмотров 4882 года назад
Understanding Fastai Learner Class Part - 2
Understanding fastai Learner Class - Part 1
Просмотров 8892 года назад
Understanding fastai Learner Class - Part 1
Deploy AI app on Jarvislabs.ai using Gradio
Просмотров 6312 года назад
Deploy AI app on Jarvislabs.ai using Gradio
I like this format of struggling through the weird stuff, because that’s what we all deal with. Very nice, thanks.
what other depth map can we use since this one doesnt work, an link recommendation?
Seems phishy.
nice video! so does running axolotl essentially make the llama3 model smarter?
One Button Prompt already does this. Reductive.
In image classification tasks, why not use focal loss instead of cross entropy for soft label data augmentation techniques such as cutmix or mixup? Why not use the sum of the focal loss for each class as follows: 'sum_i{alpha_i*y_i (1 -p_i)**gamma*log(p_i)}'?
Very interesting i listened to the very end, and it gave me some ideas for prepping my Model. Thanks for the explanation and demo. May I ask?... Do you think an Nvidia GTX Orin Devkit 64 GB would be fitting for running LLMs locally for fine tuning, training and later deploying to server once developed (both Locally and Server on Ubuntu)?
Have not tried it. No idea.
what the Home/data/img ????
Thank you very much for this video. I was having problems, but followed you along and learnt so much. Best of luck and I want to see more videos!!!
Thank you so much for your kind words! We glad the video helped you out. Your support means a lot, and we will definitely keep making more videos. Stay tuned for more content.
INT 8 for production use case ? really ?
Its already being used in some of the major companies 😀
x.com/jiayq/status/1817092427750269348
Great Video
Thank you
Phenomenal
Thank you
Good platform for renting GPUs.. pls add some discount :) I see MassedCompute giving some 50% off . I would like to use your platform
We want video on how to easily create famous personality having our product like shoes, watch,...how to do it without coding
Not sure if it is easy, we will explore. Not even sure if it is legal 😀
Possible to have the storage with a text box for input? That drag bar is so difficult to adjust. But thanks for all you release here.
Definitely yes, this is something which I hate myself. Have been asking my UI developer to prioritise. It will be done on Monday.
@@JarvislabsAI Cool!
It is done, you can enter the value in the text box.
@@JarvislabsAI Thanks wonderful
Marvellous!! Earlier I had to "keep a watch" when my workload would finish, so that I could terminate the ollama server manually.
Can’t wait to try it out. 🎉
Nice explanation
Marques😂
One on Paligemma please
Hi, sure, we will check it out and make a video about it
Thank you for this clear and concise explanation of Flux. Truly appreciate your effort!
Do you support flux.1-dev lora training?
Hi @Mr_sorus, Yes, you can fine-tune flux.dev on Jarvislabs. Soon, we will be publishing a blog on how to fine-tune the Flux model, so stay tuned with us.
Hi @Mr_sorus, we have released the video on FLux LoRA training. Please refer the link : ruclips.net/video/qiECrN0iCdI/видео.html
how to remove existing sunglasses from a face of a particular picture and swap the face with a picture of same person without sunglass from another picture of him and showing his eyes. reactor face swap not working properly on original image wearing sun shades.for movie actor arabic kuthu stills image
Inpaint the glasses and eyes into a normal face then face swap the face you want on the face that was generated after inpaint.
Hey I need help you didn't explain how u got the gemmini2 model in the if-ai node did u download it from huggingface? What folder do I need to put it in I see no Llama folder inside the models folder
from where i download the Turbo Vision Xl model?
Hi, You can find the turbo vision XL model in Civitai.
Hello, when creating a tutrial, it maybe better to show the result at the start.. Cheers :)
super ! looking forward to learning how to train an entire dreambooth :) cheers
Thanks!! Sure, we will soon drop a video on how to dreambooth train a model.
@@JarvislabsAI AWESOMEEE !
Midas depth map help link github Pleae....
Here is the repository link: github.com/LykosAI/ComfyUI-Inference-Core-Nodes. You are welcome to use any control-net preprocessors you prefer; you don't need to use the same preprocessor shown in the video
I am following this and when I execute the prompt i get an error message "prompt outputs failed validation"
Hi, If it is possible for you to share your logs with us or please ping us on the chat: jarvislabs.ai/. We're happy to help you
ComfyUI cant find the Dreamtalk node on windows :( pls help
me too,and I add nodes about IF_HFDownload,cannot show the result,and when I resatar,comfyui install python shting,I close the windows,it's worse,now comfyui always tell me it can't run ,always Error prompt.cry....pls help.
I read "Then download LLaVA's first-stage pre-trained projector weight" in the readme. Where does this go?
Hey wanna ask, does the accelerate use DDP or FSDP in default? I want to train a LLM with additional module, i froze the LLM except that additional module. Should i use FSDP?
Very nice explaination
I'm way too stupid for this
hello, do you have those workflow pic?
Hi, we have linked the workflow in the description.
is this what they call magic prompt? where ollama model refine user prompt?
The Topic is interesting but (in common with most RUclips Comfy experts, the whole presentation is confusing for the 90% of the audience that has just stumbled upon this. I think to be more successful you need to be clearer about what you want to achieve, why its a good idea. Explain how this JarvisAI fits into this, make it clear what resources need to be downloaded and exactly how in the least problematic way. I dont want to appear too negative as of course you are wanting to be helpful just trying to give some tips how to improve your presentation and hopefully consequently increase subscriber numbers
Awesome. Thanks.
Go around the world to find good comfy tutorials and land back in India ! GREAT PRESENTATION GUYS. If you take requests...can you do one on training your own dream booth. like creating old Indian comic ? cheers and subscribed. b
Sure would love to.
@@JarvislabsAI cant wait. Cheers
Its a wonderful videa, clearly and concisely explained.
Glad you liked it
Still it makes mistake and hallocinates.
They are going to make it for some more time.
@@JarvislabsAI I think it's never guaranteed...and they have already trained on Wikipedia and unreliable data like reddit,and some online blogs, websites.. it will always be useless unless u know the exact answer already.
thats why some call it a feature..
@@JarvislabsAI just to fool people and charge money for wrong information using a glorified tool
I've heard that if you use something other than 1024x1024 for your Empty Latent image node, it will then be less likely to put a watermark. I use 1016x1016.
Hi @joyceracing, we haven't tested that yet. BTW thanks for the information. We will test it out
@@JarvislabsAI your control net image had a watermark
Exactly. This is a known issue.
After two hours, still can't get Inference_Core_MiDaS-DepthMapPreprocessor added to Comfyui to use this workflow. The link associated with the missing node does not work. Where do you get this node?
Hi @DerekShenk , If you tried installing the inference core node via comfyui manager. There seems to be a download issue. No worries, you can install it manually using the following terminal commands Note: Make sure to only copy and paste the parts within the quotation marks (" ") into your terminal 1. Navigate to this file directory : (Command) "cd ComfyUI/custom_nodes" 2. Clone the repo :(command) "git clone github.com/LykosAI/ComfyUI-Inference-Core-Nodes.git" 3. Navigate into the inference core node folder: (Command) "cd /home/ComfyUI/custom_nodes/ComfyUI-Inference-Core-Nodes/" 4. Installing all the requirement : (command) "pip install ." 5. Restart the comfyui server
@@JarvislabsAI Thank you, however I had already followed those steps... git clone... and pip install. After trying three more times to install this inference node, I searched for the node and found it still failed to load when launching Comfyui. The error is " File "C:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Inference-Core-Nodes\__init__.py", line 1, in <module> from inference_core_nodes import NODE_CLASS_MAPPINGS, NODE_DISPLAY_NAME_MAPPINGS ModuleNotFoundError: No module named 'inference_core_nodes' I have updated Comfyui. Any suggestions?
Hi @DerekShenk, you might be missing some dependencies, but most of them should be downloaded by running the following commands: 'cd /home/ComfyUI/custom_nodes/ComfyUI-Inference-Core-Nodes/' and 'pip install .' Ensure that you include the dot at the end of the second command to install all the dependencies correctly. If the issue persists, you can visit our website and contact us through the chat feature
@@DerekShenk I have the same error it's wast of time!!!
You do not need the depth map they used. You can use any depth preprocessor.
thanks for the tutorial!! why the workflow operates mostly on the CPU?
Hi @Empaches, we have no idea on why that's happening. We'll check it out and let you know if we can find something to help you.
Thank you 🙂 , great video. I am working on a few updates for this but meanwhile the VHS combine node is not needed for the workflow is just to display the video, since the video gets output in the output folder. I am developing a new video previewer and quality upscaler.🎉
Oh thats great.
Awesome
@@huwhitememes thanks :D
I add nodes about IF_HFDownload,cannot show the result,and when I resatar,comfyui install python shting,I close the windows,it's worse,now comfyui always tell me it can't run ,always Error prompt.cry....pls help.
@@carterd2870 is it the IF_HFDownload? you need to connect a text display node if you are using IF_AI tools you need the last update and move the IF_AI folder from the ComfyUI-IF_AI_tools to the root of ComfyUI/input/IF_AI folder
Ollama is superslow, I would like a faster version using LM Studio or similar. Thanks
Noted!
slow or fast isn't that depends which model you are using? phi3 in ollama blazing fast
Is it available for Automatic1111??
Yes Automatic also has SUPIR
@@JarvislabsAI That's not true.
Thank you for the workflow.
How do i get the Load Checkpoint?
You can double click and search for load checkpoint node. If you want the checkpoint model. You can download it from this link: huggingface.co/RunDiffusion/Juggernaut-XL-v8/tree/main
Thank you!@@JarvislabsAI