Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
DreamRec introduced a novel 'learning-to-generate' approach to sequential recommendation where instead of the traditional method of classifying user preferences from a mix of positive and negative items, it creates an 'oracle' item that represents the ideal next choice for the user.
Sponsored
The Semantic Layer Summit is a one-day virtual event featuring the top voices in data science and analytics [12]. This year’s talks will cover recent Semantic Layer innovations for enterprise analytics, BI, and genAI. If you want to network with other leaders, learn, and get access to extra recorded content, you should register to save your virtual seat for free [12].
FlowMDM is a new model for generating long, continuous human motion sequences from text descriptions. This first-of-its-kind diffusion-based model uses Blended Positional Encodings for realistic motion creation without extra denoising steps, excelling in accuracy and realism on key datasets.
The VSP-LLM framework introduces a new approach to visual speech recognition and translation by integrating LLMs. It efficiently processes video inputs by deduplicating embedded visual features and employing Low-Rank Adaptors for cost-effective training.
This project introduces a framework for Multi-View 3D reconstruction that doesn't require prior knowledge of scene depth. It intelligently selects the most relevant source frames, making it applicable to diverse scenarios including large-scale outdoor and top-view building environments.
Spiking Neural Networks is a new efficient model for autonomous vehicles. It achieves high performance with up to 85% less energy.
San Francisco Compute Company is now offering the world's cheapest H100 training clusters with flexible short-term rental options, targeting customers needing intensive compute for AI model training without long-term commitments. Its first cluster, Angel Island, is currently online, with another, Bay Bridge, coming soon. SF Compute's model is unprecedented as it prioritizes accessibility and affordability for AI startups without demanding lengthy contracts.
Axolotl is a powerful library for fine-tuning language models.
Craft ads that click and convert.
A powerful merged and tuned model that scores well in elo and static benchmarks. It is designed for multi-turn and instruction following tasks.
Sponsored
Testing LLMs is not simple. Probabilistic output makes failures hard to identify, while running the models repeatedly tends to become very expensive quickly. In this blog post [5], QA Wolf engineer John Gluck covers 6 things the team learned about building automated black-box regression tests for genAI applications [5], including:
Microsoft is creating its own AI server hardware to decrease its dependency on Nvidia.
Google has acknowledged an issue with its AI model Gemini. The model injected inappropriate diversity into historical images, reflecting problems with bias in training data. The flaw sparked debate around diversity, equity, and inclusion in tech. Google implied that it will make improvements in the future but stopped short of a full apology for failing to properly contextualize historical figures in image generation.
OpenAI is working on a web search to compete more directly with Google. It is unclear if the product will be standalone or part of ChatGPT. Competition in the search space is filling up quickly with the addition of Copilot on Bing, newcomers like Perplexity, and Google's Gemini Copilot. A YouTube Short featuring Microsoft CEO Satya Nadella talking about competing with Google is available in the article.
REINFORCE is a simple, standard, and easily understood RL method. It is hard to train stably when used in simulators. PPO is much more performant and stable in general. Gemini uses REINFORCE and GPT-4 is believed to use PPO.
AlphaFold is used to predict the state of a protein after folding. By adding flow matching, which is invertible, you can dramatically improve modeling power on the entire landscape of proteins.
Researchers have developed a new way to make LLMs more efficient and easier to use by employing a method that focuses on 'expert-level sparsification', which reduces model size without losing performance. This is particularly useful for Mixture-of-Experts LLMs, which are powerful but usually too big to handle easily.
Sponsored
How do you transition from experimental models to highly scalable applications? apply() is a free community conference [12], where you can learn from seasoned engineers and technical leaders. Don’t miss out on the chance to hone your AI/ML skills in these FREE practical
GeneOH Diffusion is a new technique that improves how models understand and interact with objects using hands. This method focuses on making these interactions more natural by correcting errors in hand movements and relations with objects.
Snap research has trained a video generation model that reaches the previous state of the art (excluding Sora) while being 3x faster to run.
Deployed in auction bidding systems, recommendation engines, and more at Meta, Pearls is ready for research and deployment.
A model based on CodeLlama and DeepSeek Coder was able to get 85%+ on the HumanEval benchmark for programming by training on a synthetic multi-turn dataset and using human feedback.
Anthropic's research scientists have been working on a method of understanding deep neural networks that uses Circuits. These Circuits aim to identify subparts of models that get used for certain tasks. The research team has released a monthly update on the experiments they attempted and the results.
INSTRUCTIR is a new benchmark aimed at making search engines smarter in understanding users' intentions. Unlike current methods, which mostly focus on the query itself, INSTRUCTIR evaluates how well search engines can follow user instructions and adapt to various and changing search needs.
Sponsored
Join this Kolena webinar [19] to explore unique challenges in computer vision - including precise object detection and real-time decision-making - through the lens of autonomous vehicle systems. Join
This is a curated list of papers about using LLMs for Annotation.
Email-based Chief of Staff powered by AI.
Sam Altman's request for $7 trillion aims to support the rapidly escalating costs of advancing generative AI models like GPT, suggesting an exponential growth in resource needs for future iterations. This ambition underscores a pivotal moment in AI development, balancing between rapid technological progress and the broader implications of such swift advancement on safety and societal readiness.
Sponsored
Sound familiar? Your days are filled with nonstop meetings. You need every meeting to drive progress, but too often they meander with no clear outcome. Fellow ensures every meeting stays focused and impactful by creating agendas, taking notes, and following up on action items.
Stability AI has announced Stable Diffusion 3, a Diffusion Transformer similar to Sora from OpenAI. The company trained a suite of models ranging from 800m to 8B parameters, a substantial leap in size from previous image generation models. The models will be released after a period of research.
Sierra aims to revolutionize customer interactions with brands through AI agents, positing conversational AI as an essential new technology category for all aspects of customer experience.
Phind's 70B code model scores 83% on HumanEval, matching GPT-4 performance. Crucially it has significantly fewer lazy responses and refusals. The 34B model will be released soon - the larger 70B model will be released in time.
GradSafe is a new method for detecting unsafe prompts in large language models without extensive training. It can identify unsafe prompts more effectively than current methods by analyzing the gradients of specific parameters.
Class-Aware Mask-guided (CAM) feature refinement is a new method that enhances text recognition in complex scenes.
Sponsored
The third annual Semantic Layer Summit [15] is happening this April! This free, one-day virtual event features some of the top voices in enterprise analytics, BI, and genAI — including leading technical leaders at Mode, Netflix, T-Mobile, and AtScale.
This codebase allows for inference on SIMD CPU architectures, similar to llama.cpp. It was developed by Google engineers and scientists. The tool's purpose is to encourage experimentation and usage of the Gemma models released yesterday.
MMedLM 2 is a groundbreaking multilingual language model tailored for the medical field. It is trained on a new 25.5 billion-token corpus that covers six languages. MMedLM 2 outperforms other models, including GPT-4, in medical question-answering tasks.
A new method for object recognition using a language decoder. The method involves predicting text tokens from image embeddings and utilizing a specialized non-causal attention mask. It enables efficient parallel sampling of multiple labels.
Mozilla acted as a scalable evaluator for the efficiency challenge hosted last year. It ensured that entrants complied with the rules of the competition and evaluated model performance. This post is a good introduction to language model evaluation and competitive training recipes.
FunSearch, introduced in a recent Nature paper, is a method that uses large language models to make new discoveries in mathematics and computer science. It pairs a creative LLM with an evaluator to produce verifiable solutions, demonstrating success with the tough cap set and bin-packing problems. The approach yields human-interpretable solutions, marking a significant step in AI-driven scientific discovery.
Vector search company Marqo raised $12.5M in a Series A funding round led by Lightspeed to advance the adoption of its next-generation vector search platform. Marqo’s platform unlocks the massive value of unstructured data across business-critical applications like end-user search, retrieval-augmented generation, and more, helping enterprises realize the full potential of AI.
Sponsored
Learn how WellSaid Labs created a novel respelling system that enables users to turn text into more realistic, human-sounding voices