Deep Learning Weekly: Issue 420

This week in deep learning, we bring you Tencent's Hunyuan-MT translation models, Le Chat. Custom MCP connectors. Memories, and a paper on USO: Unified Style and Subject-Driven Generation via Disentangled and Reward Learning.

You may also enjoy Fine-Tuning gpt-oss for Accuracy and Performance with Quantization Aware Training, a paper on A Survey of Scientific Large Language Models: From Data Foundations to Agent Frontiers, and more!

As always, happy reading and hacking. If you have something you think should be in next week's issue, find us on Twitter: @dl_weekly.

Until next week!

Industry

Tencent open-sources Hunyuan-MT translation model series

Tencent open-sourced a new lineup of language models, the Hunyuan-MT series, that is optimized for translation tasks.

Le Chat. Custom MCP connectors. Memories.

Mistral’s Le Chat now integrates with 20+ enterprise platforms—powered by MCP—and remembers what matters with Memories.

How Sakana AI's new evolutionary algorithm builds powerful AI models without expensive retraining

A new evolutionary technique from Japan-based AI lab Sakana AI enables developers to augment the capabilities of AI models without costly training and fine-tuning processes.

MIT researchers develop AI tool to improve flu vaccine strain selection

Researchers from MIT set out to make vaccine selection more accurate by creating an AI system designed to predict dominant flu strains and identify vaccine candidates.

Anthropic triples valuation to $183B in new $13B funding round

Anthropic announced that it has raised $13 billion in funding to support its AI research and commercialization efforts.

Amazon launches Lens Live, an AI-powered shopping tool for use in the real world

Amazon launched Lens Live, a new AI-powered upgrade to its Amazon Lens shopping feature that allows consumers to discover new products through visual search.

MLOps & LLMOps

LangExtract + Milvus: A Practical Guide to Building a Hybrid Document Processing and Search System

A practical tutorial demonstrating how to build a hybrid document processing and search system by combining LangExtract for structured data extraction with Milvus.

Anatomy of a Context Window: A Guide to Context Engineering

A detailed blog post outlining the anatomy of an AI agent's context window, including system prompts, tools, memory blocks, and files, and how these components are managed.

Learn how Amazon Health Services improved discovery in Amazon search using AWS ML and GenAI

An illustrative blog post detailing how Amazon Health Services enhanced search discovery using AWS ML and generative AI.

Learning

‘World Models,’ an Old ...