Building https://predictionguard.com, creating #AI for good with SIL International, Co-host of @practicalai, aka Data Dan. #nlp / #NLProc, #artificialintelligence, #machinelearning, #datascience, #python, #golang
🦾 New episode of Practical AI!
💬 Chris and @dwhitena take a step back to look at how generative AI fits into the wider landscape of ML/AI/data science 🌄
They talk through the differences in how one approaches “traditional” supervised learning and how practitioners are approaching generative AI based solutions 🎨
Finally, they talk through the risk and compliance implications of generative AI, which was in the news this week in the EU 📰
🚀 New episode of Practical AI!
At the recent #ODSC East conference, @dwhitena got a chance to sit down with @erinmikail to discuss the process of gathering human feedback and creating an instruction tuned Large Language Models (LLM). 🗣️
They also chatted about the importance of open data and practical tooling for data annotation and fine-tuning. Do you want to create your own custom generative AI models? This is the episode for you! 🧫
It's not real until there is a LangChain integration...
PR open for Prediction Guard as an LLM in LangChain: https://github.com/hwchase17/langchain/pull/3370
Let's go!
The "Prediction Guard" system helps devs quickly find and reliably integrate the best #AI models for their use case (e.g. LLMs like those shown).
If you dig this, "like" this video by clicking 👍 on YouTube (not here): https://youtu.be/QcE2MqnAQog. I'm up for a prize at @INFORMS this week!
Wow! Excited and honored to have our @practicalai podcast featured in the Insider's list of "11 podcasts every #AI and #machinelearning enthusiast should listen to" https://www.businessinsider.com/top-artificial-intelligence-machine-learning-podcasts-2023-4
In case you missed it on the pod, @dwhitena did an excellent job walking us through co:here’s guide to prompt engineering:
Spent the weekend adding tests, database migration scripts, and some automation to the Prediction Guard (https://www.predictionguard.com/) API! Will be sending out invites to more people on the wait-list his week.
Also, I have an example in the works to show Prediction Guard and LangChain together for the best possible #AI app integrations.
LangChain:
- Prompt templates
- Chaining
- Agents
- etc
Prediction Guard:
- Model selection
- Model fallbacks
- Future proofing
- etc
Probably the most important application of #AI that I've seen recently. I'm now a saint thanks to: https://paintedsaintly.com
Wow! So honored to see the @practicalai podcast featured by Pocket Casts in their #AI revolution list! Listen to our latest episodes here: https://changelog.com/practicalai
MLOps is changing with the recent focus on foundation and generative models, but the #operations side of #AI is only increasing in importance. On this week's Practical AI Podcast, we team up with the MLOps Community podcast to discuss the changing landscape: https://changelog.com/practicalai/210
Woah, what a great post detailing 8-bit matrix multiplication for transformers at scale using Hugging Face Transformers, Accelerate and bitsandbytes! Including notebooks to run T5-11B and BLOOM in Google Colab. https://huggingface.co/blog/hf-bitsandbytes-integration
On @practicalai, @dwhitena explains RLHF, a unique model training method that is key to ChatGPT’s response quality:
Today I used Prediction Guard (https://www.predictionguard.com/) to compare 14 different SOTA machine translation models across a bunch of languages. I found an optimized model for each language pair, given a set of domain specific examples.
Long story short... NLLB from Meta/Facebook is dope. #AI #NLProc
Great to talk with @JayAlammar at #EMNLP2022, and I'm honored that he featured my team's research in his latest video! Check it out: https://youtu.be/plCvF_7qrmY
There's too much hype about Metaverse, Multiverse, and all the verses. 😀
On this week's @practicalai, Beau Perschall (NVIDIA) discusses 3D assets that are actually useful for industrial #AI applications! Simulation, synthetic data, and more. Listen now: https://changelog.com/practicalai/209
🔥Hot take: you shouldn't care about the specific #AI model you are using. It'll be obsolete in a matter of weeks.
But... there's good news. I recently built "Prediction Guard" to future proof my inferences!
🚀Launching publicly soon. Join the waitlist: https://predictionguard.com
I had a delightful time setting up a #GPU-based dev environment... When's the last time you said that? 😀
With brev.dev, I had my VS Code running stable diffusion on a GPU in the cloud in minutes. Learn more on this week's @practicalai: https://changelog.com/practicalai/208
💡 @dwhitena shares one thing he’s learned about successfully releasing ML models: https://youtu.be/AZ0oFgstqgw
Why do small organizations struggle with #MachineLearning #DataScience #AI?
Hint: It's not because they of their lack of "big data" (is that even still a term that people use?)
Find out the real reasons on this week's @practicalai with Kirsten Lum: https://changelog.com/practicalai/207
Winter break was a fun one for side projects! One of those turned into something that's changing the way I think about reliably integrating #AI into products.
"Prediction Guard" will be launching soon! 🚀
Sign up for notifications: https://www.predictionguard.com/pre-launch
