Announcing the Carnegie Mellon Database Group's Industry Affiliate Program. Thank you to our founding members: ClickHouse, Confluent, DataStax, dbt, Firebolt, Neon - Serverless Postgres, PingCAP, RelationalAI, StarTree, Weaviate • vector database. db.cs.cmu.edu/2024/09/announ…

I might've been living under a rock but I just learned about Large Action Models (LAMs) *we need a lamb emoji* ☝️ LAMs focus on pre-training language models with sequences of function calls and their actions. This makes function calling more native to the model. Zhang et al.

New model from Mistral - Pixtral 12b language vision model architecture: "dim": 5120, "n_layers": 40, "head_dim": 128, "hidden_dim": 14336, "n_heads": 32, "n_kv_heads": 8, "rope_theta": 1000000000.0, "norm_eps": 1e-05, "vocab_size": 131072, "vision_encoder": "hidden_size":

Really nice conversation w Vincent about (multi-)table retrieval (eg for dataset search, rag), and how (table) representation learning & generative models can help in surfacing insights from structured data end-to-end, efficient and reliable. Thanks for the chat Vincent D. Warmerdam!!

Superposition prompting accelerates and enhances RAG without fine-tuning, addressing long-context LLM challenges. 93× reduction in compute time on NaturalQuestions-Open with MPT-7B 🤯 Key Insights from this Paper 💡: • Parallel processing of input documents can reduce compute

Sentence Transformers v3.1 is out! Featuring a hard negatives mining utility to get better models out of your data, a new strong loss function, training with streaming datasets, custom modules, bug fixes, small additions and docs changes. Release notes: github.com/UKPLab/sentenc… 🧵

Want to keep your listings top-notch and search-friendly? Learn how to setup a real time Generative Feedback Loop (GFL) using GlassFlow.dev, Supabase, and Weaviate that brings AI-driven enhancements to Airbnb listings — ensuring data is always up-to-date and optimized for

🚀 Looking to streamline fine-tuning for models like Llama 3, Mistral, and Mixtral? Our latest blog shows how Anyscale simplifies LLMs fine-tuning with Ray: 🔑 Key takeaways: • Scalable fine-tuning with LLM-Forge • Model Deployment with RayLLM Read more:

Exciting News: FastMLX and I are Joining Arcee.ai! 🎉 This move marks an exciting new chapter in my journey to advance ML research, development and production. The Incredible Team 👨🏽‍💻 I'm honored to be part of an exceptional team of superstars at Arcee – some of the brightest

We did something with Outlines

Learn how we use Modal at Modal!

NEW SQL Console on Hugging Face Datasets Viewer🤗 🔸Run SQL on any public dataset 🔸Powered by DuckDB WASM running entirely in the browser 🔸 Share your SQL Queries via URL with others! More coming soon!

my mlops community podcast just dropped. was extremely cool to be invited. watch and learn DSPy, Ax, LLMs, and all things based

🚀New Guide on the Portkey Blog: DSPy in Production - by Ganaraj P R, based on his talk at the LLMs in Prod BLR meetup. Learn how to leverage DSPy by Omar Khattab to tackle real world challenges, optimize AI pipelines, and revolutionize e-commerce operations. Read it now:

Are you tired of having to download datasets to explore them? You can now run SQL directly in the Hugging Face Hub dataset viewer 🤗As an example, check out positive samples in the IMDB dataset huggingface.co/datasets/stanf…

🖼️ pip install -U vLLM vllm serve mistralai/Pixtral-12B-2409 --tokenizer_mode mistral --limit_mm_per_prompt 'image=4' --max_num_batched_tokens 16384 github.com/vllm-project/v…

Agent Workflow Memory Substantially improves the baseline results by 24.6% and 51.1% relative success rate on Mind2Web and WebArena while reducing the number of steps taken to solve WebArena tasks successfully repo: github.com/zorazrw/agent-… abs: arxiv.org/abs/2409.07429