MACHINE LEARNING

→ View original post on X — @datarobot

DataRobot Named 2024 Gartner Customers’ Choice for ML

By

@datarobot

–

25 February 2025 21h05

ICYMI: DataRobot was named a 2024 Customers’ Choice on Gartner Peer Insights™ for Data Science and Machine Learning Platforms. Learn more about how we’re maximizing impact for our customers: https://
bit.ly/3DJMnS2 #GartnerPeerInsights

Parallel Best-of-N Sampling and Majority Voting for LLM Improvement

By

–

25 February 2025 21h00

Sure! The simplest example would be parallel Best-of-N sampling and using majority vote or a verifier. Google had a good paper on that last summer: https://
arxiv.org/abs/2408.03314

Multi-head Latent Attention Innovation and OpenAI’s Mixture-of-Experts

By

–

25 February 2025 20h59

Good point. They brought us multi-head latent attention last summer.
(I'm not counting mixture-of-experts, because OpenAI uses it regardless of constraints :P)

Inference-Time Scaling: Prompt Modification vs Sampling and Verification

By

–

25 February 2025 20h47

*One might argue that modifying prompts (eg classic "think step by step") counts as inference-time scaling due to the extra output tokens. So in that sense they use inference-time scaling.
I should have been more clear: I meant specifically sampling techniques + using a verifier.

Compute Constraints Led to LoRA and DPO Development

By

–

25 February 2025 20h44

Compute constraints brought us LoRA and DPO

→ View original post on X — @langchain

LangSmith Adds Metadata Grouping for Experiment Analysis

By

@langchain

–

25 February 2025 20h17

📊 You can now group LangSmith experiments by metadata to get valuable insights!

We’ve added a new view that allows you to group experiment results by metadata. Compare the performance of your evaluations across different segments (eg. user segments or subject areas) to pinpoint… pic.twitter.com/2nGo2bm7JY
— LangChain (@LangChain) 25 février 2025

You can now group LangSmith experiments by metadata to get valuable insights! We’ve added a new view that allows you to group experiment results by metadata. Compare the performance of your evaluations across different segments (eg. user segments or subject areas) to pinpoint

→ View original post on X — @ludosln

Two AIs, ChatGPT and Claude 3.7, control a computer and work solo

By

@ludosln

–

25 February 2025 19h58

I connected #ChatGPT and the new #AI Claude 3.7 together… And then, I saw the future These 2 AIs took control of my computer and worked for me… Without me! It went very far, watch my video → https://
youtu.be/bMoHStBFLcI

→ View original post on X — @nvidia

Amphitrite Uses AI Satellite Data for Ocean Current Prediction

By

@nvidia

–

25 February 2025 19h34

Amphitrite, a member of the #NVIDIAInception program, uses satellite data and #AI to simulate and predict oceanic currents and weather. ➡️ https://t.co/nN0JAwAN54

This helps reduce travel times, fuel consumption, and, ultimately, carbon emissions. pic.twitter.com/KWCkuS7g4V
— NVIDIA (@nvidia) 25 février 2025

Amphitrite, a member of the #NVIDIAInception program, uses satellite data and #AI to simulate and predict oceanic currents and weather. https://
nvda.ws/4ico3XT This helps reduce travel times, fuel consumption, and, ultimately, carbon emissions.

→ View original post on X — @genekogan

pix2pix: Understanding GANs with Image Input Architectures

By

@genekogan

–

25 February 2025 17h53

pix2pix was actually a GAN! Just one with an image input. as usual, there are more caveats and exceptions than rules. https://
arxiv.org/abs/1611.07004 the distinction i've always kept in my head was that the "generative models" are such that the model (i.e. the neural net) directly