Author

Machine Learning Algorithms & Natural Language Processing

Focused on frontier AI technologies, empowering AI researchers' progress.

586

Articles

Likes

2.9k

Views

Comments

Latest from Machine Learning Algorithms & Natural Language Processing

100 recent articles max

Machine Learning Algorithms & Natural Language Processing

Aug 29, 2026 · Artificial Intelligence

Dropping Intermediate Tokens: How Prefix Sliding Achieves Up to 3× Faster Long-Context Reasoning

Prefix Sliding keeps the task prefix and a sliding window of recent tokens while evicting older intermediate tokens from the KV cache, enabling up to three‑fold speedups for long‑chain inference without retraining and extending reinforcement‑learning rollouts beyond 100 k tokens.

AttentionEfficiencyKV Cache

0 likes · 11 min read

Dropping Intermediate Tokens: How Prefix Sliding Achieves Up to 3× Faster Long-Context Reasoning

Machine Learning Algorithms & Natural Language Processing

Aug 29, 2026 · Artificial Intelligence

Can Chinese‑Only Inference Training Match English? Apple’s New Study Shows Only 1.1% Gap

Apple and the Hasso Plattner Institute evaluated over 200 multilingual inference training experiments across nine base models and eleven languages, finding that training with Chinese rewards incurs just a 1.1 percentage‑point loss versus English, while low‑resource languages can cause severe performance collapses in specific model‑language combos.

GemmaQwen3cross-language transfer

0 likes · 9 min read

Can Chinese‑Only Inference Training Match English? Apple’s New Study Shows Only 1.1% Gap

Machine Learning Algorithms & Natural Language Processing

Aug 28, 2026 · Artificial Intelligence

Visual Tracks: A New Language for Robot World Models (TrAct)

The paper introduces TrAct, which replaces action‑conditioned world models with visual‑track conditioning, letting a policy output both robot actions and 2‑D visual trajectories that guide a future‑prediction model, and demonstrates substantial gains on the LIBERO‑INTEGRAL benchmark and real‑robot tests.

TrActbenchmarkembodied AI

0 likes · 9 min read

Visual Tracks: A New Language for Robot World Models (TrAct)

Machine Learning Algorithms & Natural Language Processing

Aug 28, 2026 · Artificial Intelligence

Code-as-World: Reconstructing the Physical World from Video

The article introduces Code-as-World, a framework that turns visual observations into executable, verifiable code describing objects, dynamics, and appearance, enabling agents to iteratively propose, simulate, and refine world hypotheses and achieving state‑of‑the‑art results on physical reasoning benchmarks.

Agentic DiscoveryCode-as-WorldExecutable World

0 likes · 13 min read

Code-as-World: Reconstructing the Physical World from Video

Machine Learning Algorithms & Natural Language Processing

Aug 27, 2026 · Operations

Why Removing the Vision Encoder Improves Infrastructure Efficiency

The article analyzes how eliminating the Vision Encoder from multimodal large language models simplifies the computation graph, reduces load imbalance, avoids conflicting parallel configurations, and makes scheduling and performance modeling more predictable, especially when visual features can be pre‑computed offline.

Cost ModelingEncoder-freeMultimodal LLM

0 likes · 9 min read

Why Removing the Vision Encoder Improves Infrastructure Efficiency

Machine Learning Algorithms & Natural Language Processing

Aug 26, 2026 · Artificial Intelligence

Can AI Do Independent Research? ASI‑Bench Measures Scientific Autonomy

ASI‑Bench, developed by Tsinghua and leading institutions, is a benchmark that evaluates AI’s scientific autonomy by progressively reducing method guidance across four levels, revealing that current models lose up to half their scientific score without detailed instructions, highlighting the gap to true independent research.

AI autonomyASI-Benchagent evaluation

0 likes · 14 min read

Can AI Do Independent Research? ASI‑Bench Measures Scientific Autonomy

Machine Learning Algorithms & Natural Language Processing

Aug 26, 2026 · Artificial Intelligence

From Experience to Ability: How Agentic Skills Form, Internalize, and Self‑Evolve

In this MLNLP Academic Talk, Tsinghua PhD candidate Wu Jinyang presents his research on Agentic Skill formation, internalization, and continual self‑evolution, detailing three projects—ThoughtICR, TemplateRL, and SEED—that connect contextual reasoning, reinforcement learning, and autonomous skill growth.

Agentic SkillSEEDTemplateRL

0 likes · 4 min read

From Experience to Ability: How Agentic Skills Form, Internalize, and Self‑Evolve

Machine Learning Algorithms & Natural Language Processing

Aug 25, 2026 · Artificial Intelligence

When Online Distillation Goes Off‑Track: Relay‑OPD Lets the Teacher Take the Baton

The paper identifies the “prefix failure” problem in on‑policy distillation, proposes Relay‑OPD—a handoff‑triggered, budget‑controlled teacher takeover mechanism that corrects early mistakes, and demonstrates across eight math‑reasoning benchmarks that it improves accuracy by up to 7.3% while halving training trajectory length.

Relay‑OPDlarge language modelsmath reasoning benchmarks

0 likes · 13 min read

When Online Distillation Goes Off‑Track: Relay‑OPD Lets the Teacher Take the Baton

Machine Learning Algorithms & Natural Language Processing

Aug 25, 2026 · Artificial Intelligence

Understanding Multi‑Round Rollouts, Context Reconstruction, and RL Training in Agentic RL

The article analyzes how Agentic RL decouples internal state, protocol requests, and token sequences, explains the inference pipeline, the challenges of preserving prefix relationships across multi‑round rollouts, and details a gateway‑based data collection and credit‑assignment pipeline for reinforcement‑learning training.

Agentic RLPiRL training

0 likes · 13 min read

Understanding Multi‑Round Rollouts, Context Reconstruction, and RL Training in Agentic RL

Machine Learning Algorithms & Natural Language Processing

Aug 24, 2026 · Artificial Intelligence

Can LLMs Engineer Their Own Infrastructure? A Deep Dive into Φ‑Bench’s Assessment

This article examines Φ‑Bench, a comprehensive LLM infrastructure benchmark that evaluates how well large language models can perform real‑world infra engineering tasks, revealing current models’ strengths, weaknesses, and the gap to becoming true AI engineers.

AI engineeringError AnalysisInfrastructure Benchmark

0 likes · 12 min read

Can LLMs Engineer Their Own Infrastructure? A Deep Dive into Φ‑Bench’s Assessment