Daily digest
4 items · ~4 min · Week 2026-W31
Worth knowing (3)
Putin signs law establishing legal framework for large foundational AI models in Russia
On July 26, 2026, Russian President Vladimir Putin signed legislation creating a legal framework for developing, deploying, and applying large foundational AI models in Russia. The law introduces two model categories — "sovereign" (fully developed and controlled by a Russian legal entity, hosted in Russian data centers) and "national" (developed by a Russian entity but may include foreign open-licensed components) — both requiring at least one billion parameters and compliance with Russian law.
Molt: A Scalable PyTorch-Native Training Framework for Agentic Reinforcement Learning
NVIDIANVIDIA open-sources Molt, a PyTorch-native asynchronous RL training system built to stay small and readable enough for a single researcher to hold in their head, while matching the throughput of Megatron-based stacks. It supports multimodal and mixture-of-experts policies and enforces on-policy training by never training on a token the current policy did not generate. Uses Ray for placement/async queues, vLLM for rollout, and NVIDIA AutoModel + FSDP2 for training.
AREX: Towards a Recursively Self-Improving Agent for Deep Research
AREX is a family of deep-research agents (4B dense and 122B-A10B MoE) trained with reinforcement learning on verified synthetic tasks. It runs a dual-loop process: an inner loop gathers evidence and drafts provisional answers, an outer loop audits the answer constraint-by-constraint and dispatches targeted follow-up searches for unresolved claims. A learned context-update tool autonomously compresses growing interaction history into a compact state that preserves verified evidence.
For reference (1)
K12-KGraph: A Curriculum-Aligned Knowledge Graph for Benchmarking and Training Educational LLMs
The authors build K12-KGraph, a knowledge graph extracted from Chinese K-12 textbooks (math, physics, chemistry, biology), and derive two resources from it: K12-Bench, a 23,640-question benchmark spanning five task categories testing curriculum and prerequisite understanding, and K12-Train, a fine-tuning dataset with text and visual question-answer pairs. Evaluation shows current LLMs struggle with prerequisite and concept-relationship reasoning, while domain-specific fine-tuning on K12-Train substantially closes the gap.