"Meta-Harness: End-to-End Optimization of Model Harnesses" A big portion of agentic LLM performance relies on human-designed harness around the model, not just the weights. On top of that, how well a harness is designed could impact a model's performance heavily. So this paper
@askalphaxiv
-
Memory Sparse Attention Scales Models to 100M Token Context
By
–
Scaling Attention to 100M context!? Memory Sparse Attention introduces an idea where instead of rereading an entire 100M-token entry, it learns to jump straight into the relevant memories and reason from them end-to-end. More specifically, it first encodes documents into
-
Scientific Article arXiv 2603.25040 from March 30 2026
By
–
read more: https://alphaxiv.org/abs/2603.25040
-
Intern-S1-Pro: Trillion-Parameter Multimodal Scientific Foundation Model
By
–
"Intern-S1-Pro: Scientific Multimodal Foundation Model at Trillion Scale" Intern-S1-Pro scaled a multimodal model to 1T parameters with a lot of aligned scientific data, obtaining a really strong model that's capable of analyzing scientific figures, reason across STEM topics,
-
Natural-Language Agent Harnesses Improve Model Performance Architecture
By
–
“Natural-Language Agent Harnesses” This paper argues that agent performance increasingly depends on the harness around the model, but that harness logic is usually buried in controller code and runtime-specific conventions. So they propose Natural-Language Agent Harnesses
-
Self-Distillation Hidden Layers Self-Supervised Vision Models
By
–
"Self-Distillation of Hidden Layers for Self-Supervised Representation Learning" This paper shows that self-supervised vision models would work much better when they predict a teacher's hidden layers across the entire visual hierarchy. As they showed that just learning from the
-
Self-Distillation Degrades LLM Reasoning Capabilities
By
–
"Why Does Self-Distillation (Sometimes) Degrade the Reasoning Capability of LLMs?" Self-distillation can make LLMs look smarter by producing shorter, more confident reasoning traces, but in math it often takes out the model's uncertainty and self-correction signals. This can
-
RL Training for Distributional Reasoning in Language Models
By
–
"Reaching Beyond the Mode: RL for Distributional Reasoning in Language Models" Instead of standard RL post-training collapsing an LLM toward one dominant answer, this paper shows you can train it to produce a set of plausible answers in a single pass. This is important because
-
Language Models Drive Novel Scientific Discovery Beyond Benchmarks
By
–
LLMs aren't just chatbots, they can also search for novel discoveries! In this AI4Science talk, Yuanqi Du (
@YuanqiD
) walks through a shift in how to think about language models in science. Instead of asking whether they “understand” science through benchmarks or exams, the work -
Scientific Article arXiv 2603.24511 Available for Reading
By
–
read more: https://alphaxiv.org/abs/2603.24511