Fine-Tuned: January 2024

January 28, 2024 · less than a minute read
Newsletter LP Banner January

Make sure to sign-up for our monthly newsletter to receive all the latest AI and LLM news, tutorials and product updates.

We hope your 2024 is off to a great start! 2023 was certainly transformative for AI but we think 2024 will see shifts that make AI even more accessible, reliable, and secure for practitioners. To that end, we’re excited to release our AI and LLM Predictions for 2024, and to announce our first webinar of 2024! Read on for more details.


Featured Event

Fine-Tuning Zephyr-7B to Analyze Customer Support Call Logs

Join us on February 1st at 10:00 am PT to learn how you can leverage open source LLMs to automate one of the most time consuming tasks of customer support: classifying customer issues. You will learn how to efficiently and cost-effectively fine-tune an open source LLM with just a few lines of code at a fraction of the cost of using a commercial LLM and how to easily implement efficient fine-tuning techniques like LoRA and quantization.

Featured Event

Recent Events + Podcasts

LLMOps for Your Data: Best Practices to Ensure Safety, Quality, and Cost
WEBINAR

LLMOps for Your Data: Best Practices to Ensure Safety, Quality, and Cost

Predibase CTO Travis Addair joined Shreya Rajpal, Co-Founder and CEO at Guardrails, to discuss how to work with LLMs in ways that mitigate risks, improve the development process, and better leverage your data as an asset.

Read full story
PODCAST

TFiR “Let’s Talk”: Predibase Makes It Easy For Anyone To Train LLMs

Predibase CEO Devvret Rishi joins the TFiR “Let’s Talk” show to share how developers can fine-tune open-source models with their own data to achieve significant cost savings and performance improvements compared to commercial LLMs.

Watch
PODCAST

Caveminds Podcast: Deploy Faster, Cheaper, Smaller AI Fine-Tuned Models

The discussion explores Predibase’s unique approach to AI, focusing on its ability to streamline complex AI pipelines into concise, efficient code.

Watch
Crafted | The Artium Podcast: Taking GenAI From Prototype to Production
PODCAST

Crafted | The Artium Podcast: Taking GenAI From Prototype to Production

This episode of “Crafted” is a recording of a panel Predibase Staff Engineer Justin Zhao spoke on at NYC Tech Week 2023 and covers how to take generative AI from prototype to production, how to overcome the challenges of evaluation and performance, and more.

Read full story

Featured Blog Post

AI and LLM Predictions for 2024

From the coming wave of small language models to the future of fine-tuning and LLM architectures, these predictions represent the collective thoughts of our team of AI experts with experience building ML and LLM applications at Uber, AWS, Google, and more.

Featured Blog Post

Customer Spotlight

Deep Learning on Snowflake: How Paradigm Built a Personal Trading Experience with Predibase

Paradigm, one of the world’s largest institutional liquidity networks for cryptocurrencies, was able to use Predibase to build a deep learning-based recommendation system for their traders on top of their existing Snowflake Data Cloud with just a few lines of YAML. Building production models on top of Snowflake data–a task that used to take months–now takes minutes.

Customer Spotlight

From the Community

Top AI startups poised for success in 2024: a Roundup of Innovation
Roundup

Top AI startups poised for success in 2024: a Roundup of Innovation

Predibase is honored to be listed as one of Go Global World’s Top 10 AI startups for 2024. Read more about us and the others on the list!

Read full story
Fine-Tune Mistral with Ludwig - Demo
Demo

Fine-Tune Mistral with Ludwig - Demo

Noah Gift walks through a quick 3-minute demo of fine-tuning Mistral with Ludwig. He describes it as “a pretty strong contender for the simplest possible way to do fine-tuning” and we couldn’t agree more!

Read full story
O’Reilly Course: Responsible Generative AI and Local LLMs
Course

O’Reilly Course: Responsible Generative AI and Local LLMs

Noah Gift and Alfredo Deza recently released “Responsible Generative AI and Local LLMs,” a video series that covers using local LLMs and frameworks like Mistral, llamafile, Candle, and LoRAX.

Read full story
Coding for non-coders
Community Blog

Coding for non-coders

This article provides an overview of declarative ML and highlights a number of common challenges teams face when developing ML systems, and takes a look at fine-tuning LLMs with Ludwig.

Read full story
30 AI Libraries For The Modern AI Stack
Roundup

30 AI Libraries For The Modern AI Stack

The AI Engineer–a community focused on AI for hackers–recently released their first list of 30 open-source libraries that simplify building AI and we’re thrilled to see Ludwig included on the list.

Read full story

Open Source Updates

Tutorial: How to Fine-tune Mixtral 8x7b with Open-source Ludwig

Mixtral 8x7B is one of the first successful open-source implementations of the Mixture of Experts architecture (MoE) and we’ve made it easy to fine-tune for free on commodity hardware using Ludwig—a powerful open-source framework for highly optimized model training through a declarative, YAML-based interface. Follow along with this hands-on tutorial!

Open Source Updates

Ludwig v0.9.2 & v0.9.3: These releases introduces a few bug fixes and several new enhancements:

  • Add support for the official microsoft/phi-2 model
  • Ensure correct padding token for Phi and Pythia models
  • Cast LLMEncoder output to torch.float32, freeze final layer at init
  • Enable IA3 adapters
  • Add batch size tuning for LLMs
  • Per-step token utilization to tensorboard and progress tracker
  • Default LoRA target modules for #Mixtral and #Mixtral-instruct
  • Support for exporting models to Carton (thank you Vivek Panyam!)

LoRAX v0.6: The latest release adds support for multi-turn chat conversations with dynamic LoRA adapter loading. Just replace the "model" parameter with any HF LoRA and you're set. Chat templates can come from the base model, or even LoRA itself if it uses its own custom template. All of this happens dynamically per request.

Featured Product Update

We’re excited launch our new prompting experience in the Predibase UI which allows you to easily prompt serverless endpoints and your fine-tuned adapters without needing to deploy them first. This lets teams test their fine-tuned models and compare model iterations all from the UI, enabling much faster test and review cycles.

Prompt-UI-2

Related Articles

Join Our
Slack Community!