- 645
- 3 392 996
Weights & Biases
United States
Registrace 17. 07. 2018
Weights & Biases helps AI developers build better models faster. Quickly track experiments, version and iterate on datasets, evaluate model performance, reproduce models, and manage your ML workflows end-to-end.
Please subscribe to our channel and turn on notifications for future videos!
Please subscribe to our channel and turn on notifications for future videos!
Robotics AI for Industrial Applications
👀 *All of the Fully Connected San Francisco 2024 videos are available at wandb.me/fcsf24yt*
*About Kyle Coelho and Brian Zhu's Session: Robotics AI for Industrial Applications*
In this session from the Fully Connected conference, Siemens' research engineers Kyle Coelho and Brian Zhu share their expertise on applying AI to robotics for industrial applications.
From computer vision to reinforcement learning, discover how these technologies are enabling automation in manufacturing and logistics, addressing both technical challenges and economic motivations. Watch the video to learn more about the transformative potential of robotics AI.
*About Kyle Coelho and Brian Zhu's Session: Robotics AI for Industrial Applications*
In this session from the Fully Connected conference, Siemens' research engineers Kyle Coelho and Brian Zhu share their expertise on applying AI to robotics for industrial applications.
From computer vision to reinforcement learning, discover how these technologies are enabling automation in manufacturing and logistics, addressing both technical challenges and economic motivations. Watch the video to learn more about the transformative potential of robotics AI.
zhlédnutí: 396
Video
Insights into Building a High-Performance Computing Cloud Platform
zhlédnutí 161Před 9 hodinami
👀 *All of the Fully Connected San Francisco 2024 videos are available at wandb.me/fcsf24yt* *About Rahul Talari and Harsh Banwait's Session: Insights into Building a High-Performance Computing Cloud Platform* Discover the insights behind building a high-performance computing cloud platform in this session from the Fully Connected conference. Rahul Talari, Senior Machine Learning Engineer, and H...
Samba-1, Enterprise Grade Open Source AI
zhlédnutí 362Před 12 hodinami
👀 *All of the Fully Connected San Francisco 2024 videos are available at wandb.me/fcsf24yt* *About Anand Sampat's Session On Enterprise Grade Open Source AI* Explore the transformative potential of Samba One in enterprise AI with Anand Sampat, Research Engineer at SambaNova, during his presentation at the Fully Connected conference. Anand discusses how SambaNova's cutting-edge hardware and plat...
Enterprise Model Management features
zhlédnutí 29Před 12 hodinami
Tune in to Hamel and Noa sharing some tips on how to take advantage of enterprise-grade model management: -Using external files docs: docs.wandb.ai/guides/artifacts/track-external-files?#docusaurus_skipToContent_fallback -Protected aliases docs: docs.wandb.ai/guides/model_registry/access_controls?#add-protected-aliases
Automation design patterns
zhlédnutí 25Před 12 hodinami
In this lesson Hamel compares Webhook vs. Launch automation and provide guidance on when to use them.
Creating a Launch automation
zhlédnutí 69Před 12 hodinami
In the previous lesson, we created the evaluation job and triggered it manually. Now lets hook it up to the Model Registry.
LLM Evaluation results
zhlédnutí 64Před 12 hodinami
In this lesson we will take a look at the results from our automated evaluation runs and make conclusions about our candidate model.
Setting up Launch
zhlédnutí 38Před 12 hodinami
In the previous chapter, you've learned from Hamel how to run your automation through Weights & Biases Model Registry and Webhooks. In this session, we will focus on running our automation through Weights & Biases Launch. We will use the following parts of the course repo: github.com/wandb/edu/tree/main/model-management -Dockerfile.eval file: github.com/wandb/edu/blob/main/model-management/Dock...
Setting up LLM evaluation
zhlédnutí 57Před 12 hodinami
After training a model, we need to understand how good it is. We will use LLM as a judge method for the evaluation of our models. In this session we will use the following parts of the course repo: github.com/wandb/edu/tree/main/model-management eval.py code: github.com/wandb/edu/blob/main/model-management/eval.py Further resources: -LLM Engineering: Structured Outputs course with Jason Liu: ww...
Finetuning an LLM and saving model
zhlédnutí 103Před 12 hodinami
Before we can manage our models, we need to train some! In this session we will use the following parts of the course repo: Train.py code - follow Darek and train your own model: github.com/wandb/edu/blob/main/model-management/train.py Data.py code: github.com/wandb/edu/blob/main/model-management/mini_llm/data.py Further resources: Training and Finetuning LLMs course: www.wandb.courses/courses/...
Introduction to Launch
zhlédnutí 45Před 12 hodinami
Easily scale training runs from your desktop to a compute resource like Amazon SageMaker, Kubernetes and more with W&B Launch. Once W&B Launch is configured, you can quickly run training scripts, model evaluation suites, prepare models for production inference, and more with a few clicks and commands. Find out more in Launch docs or connect with W&B Launch team. docs.wandb.ai/guides/launch wand...
LLM case study overview
zhlédnutí 27Před 12 hodinami
We welcome another Weights & Biases guest instructor, Darek Kleczek! Darek is a Machine Learning Engineer at Weights & Biases and a Kaggle Competitions Grandmaster. Darek will introduce a case study which will allow us to experience model management and automations while finetuning and evaluating a Large Language Model. Resources: Training and Finetuning LLMs course with Jonathan Frankle (Mosai...
Webhook Exercises
zhlédnutí 27Před 12 hodinami
It is time for you to get your hands dirty, there is no better way to learn than to try it. Here are some exercises I recommend to start with but you can get as creative as you want.
Creating Webhook automation
zhlédnutí 31Před 12 hodinami
Now that the webhook is set up and tested we are moving to setting up the automation in the model registry. Resources: Create a webhook automation docs: docs.wandb.ai/guides/model_registry/automation?#create-a-webhook-automation
Continuous Deployment with Weights & Biases Automations
zhlédnutí 90Před 12 hodinami
Continuous Deployment with Weights & Biases Automations
Human x Machine: From Models to Products
zhlédnutí 338Před 21 hodinou
Human x Machine: From Models to Products
How Mercari is using Gen AI to define the future of Japanese C2C e-commerce
zhlédnutí 337Před 21 hodinou
How Mercari is using Gen AI to define the future of Japanese C2C e-commerce
Fireside Chat with Lukas Biewald and Manuvir Das
zhlédnutí 210Před dnem
Fireside Chat with Lukas Biewald and Manuvir Das
The Future of AI in Coding with Codeium CEO Varun Mohan
zhlédnutí 1,2KPřed dnem
The Future of AI in Coding with Codeium CEO Varun Mohan
Understanding LLM Performance in Snowflake Using Weights & Biases And Snowpark Container Services
zhlédnutí 1KPřed dnem
Understanding LLM Performance in Snowflake Using Weights & Biases And Snowpark Container Services
thankyou
Yeah this video was predated 2020 collective effort on Twitter of initial working AI model that went viral for free? Each year that goes by hierarchy madness gets worse and worse because the people unwittingly greedily allowed an anonymous donor of methodology to dethrone all hierarchy all at once meaning king of kinds. Sad pathetic depraved people roam this earth looking for guidance it's not that hard to make a splash. Due respects all hail the peoples king the rest of you infinity print is for biggest cover up ever lies begit lies your gonna end up going to war with your children you know that right straight impass they think your crazy you know their crazy becauseyou did it to them letting crazy people lead never filling you in all you ask for is guidance if anyonehas to rule.
My favorite programming language is English
The ocaml user is clearly janestreet
Great questions 👍
Where is the data going to come from ? There are strict hippa regulations especially in USA
thank you so much for your help. i was able to load my custom YOLOv5 model on windows and it works perfectly. Im extremely grateful to you!
Great discussion. Concrete and accessible.
lol he doesn't give a fuck and it's awesome
Good Session.
Very interesting. I think the media hype is moving faster than the progress of AI's.
Fake accent is irritating can't focus what she is speaking
This is how modern highly educated Indians speak , she is not faking her accent , go and check any Indian English news channels for reference
@@nagendrasrinivas-cj7sr They speak like yam, yan, Ohho, yas, I work with many and they speak shit.
Great Talk. Indeed insightful as well.
LLM-based coding assistants are especially nice to those of us who habitually over-comment. My preferred programming language is .... English!
breath of fresh air, great talk
a lot of times, we want to believe agents will not replace the coder jobs it WILL replace the "coder job" that fills up rows of cubicles at large companies coders that feed into large bureaucracies of self-important software maintenance teams writing the same database/web application the same way as the last decade I think the actual coder jobs (at the front edge of software) and the new AI-native coder jobs are the things we want to protect and conserve
Weave can let you turn your inference budget into reliable data. 100% WandB = model insights
Hi, this video is HELPFUL and RATED 5/5 by me. But I wonder can I use custom dataset instead of the pre-set dataset as shown?
yes
Nice session
Thank you very much for this tutorial. I was easily able to follow and implement all of the steps. I'm really grateful to you
Excellent talk on Nvidia stack. The demo on NIM was good.
Great talk
Also...first!
Great talk. Really like the insights into the changes required as ML moves towards software developers.
Best tracking solution for years now. Keep it up
Where’s the rest of the video? 😢
Here 🙂 czcams.com/video/Z4D6l_Kc7vI/video.html
The brain is not the mind The brain is not the mind The brain is not the mind Demis gonna win #DeepMind #EZ
Thanks for this. Professor Finn is a legend. I’ve been working through CS330.
Excellent questions by Lucas. Insighful discussion.
Should saturated benchmark be retired?
Guariscimi l aids l oids e u e la sindrome psichiatrica mi chiamo Biagio di balsamo
I fucking hate this guy. So full of himself and trying to be humble.
.. becoming comfortable with being uncomfortable ❤️
This is awesome.
❤
Eenneeabbaannbbaeeaaaeebbaabnnnaneebaaeebeeaeeebbaaannannabaeeannbaaaaabbbnnneeeeaeeeebbnnneeeebnnebbeeaeeeeeebaabaaannbbbbeee
Qreerbbeeqnrrerbbeeeeeeqreeebbbbeeeeebeeeeeeeqnnrnnnreebnqeebeeeeeeeebeeeeeebnnnnbeeeeeebbqeebqannbbeebbeebaaeeabeebannnqbbbeeeebaeeeeeaeebebaeannanneebaabaannaabebbaeaeeeaeebqeebannannaannabbannannbnnaannbannbeeeebnnnnnneabnqeeebaeebnaaaaaarrnnee
As mentioned in a previous comment, a significant challenge in applying machine learning to drug discovery projects lies in the scarcity of robust and well-structured data. For instance, a major factor contributing to the failure of drug discovery endeavours is the suboptimal ADMET (absorption, distribution, metabolism, excretion, and toxicity) properties. The landscape could be transformed if we could develop models capable of predicting the outcomes of in vitro assays, allowing us to streamline the selection of well-optimized candidates for pre-clinical trials. However, the publicly available ADMET data is notably deficient in both quality and quantity, leading to the development of models that lack robustness.
Summary: Safety and size. The end.
Also 22:08 commenting on Lukas' question. The data in biological world are different from NLP or CV data in various ways, just to name a few: 1. In biology, the experiment data is only an estimation of the physical ground truth and often inconsistent, whereas in many other domain basically the test corpus used for model training is the same in training and real world. So the intrinsic noise within would impact the ceiling of how a model could be evaluated. Since the data is not ground truth, there is a greater gap between model output and reality, given even if the model is perfect on the testing data. 2. The lack of data is real. Partially because bio data is expensive. For CV an annotator could label a dozen or even a hundred pictures per hour and it costs less than $100. But in bio world, on average a single row of data could cost $100-$1000, even over $10k or more for things like protein structure, and takes days or weeks generate. It also requires high level expertise to conduct these experiments, and often repeats need to be done to analyze the intrinsic variances of these data. 3. The format of bio data is so diverse. For LLM, text is all you need, add voice and moving pictures we can train SORA. But in biology, there are hundreds of tasks, structure, affinity, stability, toxicity... each task has many different experiment types. Well. If you are interested in more about this my twitter is also NachuanShan. I work at BioMap as a data product manager, building protein language models.
20:27 cyberattackers watching this: "wtf I love ChatGPT now"
Great conversation. Love this topic.
What will be the SQLite of LLMs, with capability for local use? Llama?
Very insightful and informative
How silly is to redteam a model which you control the training data to check for bioweapons capabilities. How stupid should you have to be? Isn’t easier to run a search on the data 😂😅
How much did it cost to build, including hardware and engineering costs?
vin diesel!
There's a universe where Joseph Spisak is Mark Zuckerberg's brother. Oh, and nice presentation. Wonderful work they are doing at Meta AI.
My favorite fact from this is that the smarter the model, the more it violates rules. Just like us :)
Very true ! People who are way smarter on tax laws are the one who violate most , innocent people pay more than what they are supposed to etc . Same goes with many other laws
Or, the rules it uses instead of the rules we assumed are different.
That's a great way to justify corruption and awful people.
@@why.do.I.even.try. awful people are still human, best to understand how good people become awful
@@Crux69 Yes but we shouldn't repeat their actions just because they work. We should work towards more ethical means to advance, technologically and societally.
Congratulations!