Latest

Every published item, in order · 14 items

Nov 6

Papers US vs ChinaMulti-sourceSelected00:32

MiMo-V2.6-Pro Tops AA-PersonalityBench in Question-Ending Responses

Response-ending behaviour shapes how conversational and proactive an assistant feels, and this benchmark turns that personality trait into a measurable signal. Product teams choosing a model for customer-facing chatbots can use these figures to match tone, since a 34% question rate reads very differently from a 0.1% one. The spread between MiMo-V2.6-Pro and DeepSeek V4.1 Flash also shows two Chinese labs optimising for opposite interaction styles.

#MiMo-V2.6-Pro #DeepSeek #GPT-6 #AA-PersonalityBench

Event

官方账号Artificial Analysis@ArtificialAnlys11 sourcesSource

Nov 3

Papers BuildersMulti-source14:12

Chinese contentSSM 对角等价性研究：统一 Mamba 1-3 并提出 SSD-MIMO 检索扩展

一篇把 Mamba 1 到 3 统一进同一数学框架的理论论文，还附带更快的 GPU 核和新的 MIMO 检索扩展，适合研究序列模型的读者。

#Mamba #SSM #线性注意力 #GPU kernels

Event

arXiv cs.LG@Dennis Rochau4 sourcesSource

Papers Builders14:11

Chinese content基于图神经网络的Cell-Free MIMO通感一体化波束成形框架

一篇通信方向的论文：用图神经网络做通感一体化波束成形，单个模型能同时扛住信道误差和杂波，速率保住各专用基线的约97%。

#GNN #ISAC #MIMO #波束成形

arXiv cs.LG@Oways Alsoloh et al. (5)Source

Oct 2

Papers BuildersSelected10:41

Multi-Token Prediction Heads Research

This research significantly reduces the computational cost of implementing MTP heads, making advanced inference optimization more accessible. The findings could enable faster deployment of large language models without requiring massive pretraining resources. Developers and researchers working on model optimization can implement these techniques to improve throughput with minimal computational overhead.

#Qwen3 #MiMo-7B #multi-token prediction #inference optimization

arXiv: DeepSeek@Prachi Badarayani et al. (11)Source

Sep 24

Papers BuildersOfficialSelected10:01

Xiaomi Details HySparse2 Sparse Attention for MiMo-V3

Two-level KV sharing targets the long-context prefill bottleneck that agentic workloads hit, cutting compute without a full attention redesign. Engineers building long-context inference stacks can benchmark the approach against hybrid SWA baselines on similar MoE setups. The MiMo-V3 linkage also signals how Xiaomi positions its model stack against domestic rivals like DeepSeek and Qwen.

#Xiaomi #MiMo-V3 #HySparse2 #sparse attention

Event

官方一手pandaily@[email protected] (Pandaily)3 sourcesSource

Sep 23

Papers BuildersSelected73°04:40

ReFigBench shows harness choice changes coding agent results

The same model can score differently depending on the harness around it, so benchmark numbers reflect the full stack rather than the model alone. Teams evaluating coding agents should rerun comparisons under their own harness before drawing conclusions. ReFigBench adds a new evaluation axis, and details are on arXiv (2609.18844) alongside a chat interface from dair.ai.

#ReFigBench #GPT-5.5 #Claude Code #Codex

elvis@omarsar0Source

Sep 21

Papers Builders13:56

Chinese contentCodeMidas：以源代码为输入自动构建编码智能体 RL 训练环境

把现成开源代码库自动变成智能体 RL 练习题，MiMo-V2.5 训完 ProgramBench 涨了 17%。

#CodeMidas #MiMo-V2.5 #GRPO #强化学习

arXiv cs.AI@Bowen Ye et al. (19)Source

Sep 17

AITOP16:09

Anthropic拟10月在新加坡设亚太第五个办事处开...

Anthropic拟10月在新加坡设亚太第五个办事处开放本地招聘数据显示，按人口比例计算，新加坡的Claude.ai使用水平在121个国家中排名第二。我们观察到，人工智能公司Anthropic计划于10月在新加坡设立其亚太地区的第五…

AITOPRead

Papers BuildersSelected12:10

Interpretable Multi-Instance Learning for Early AML Prediction

This research demonstrates a practical method to predict critical AML mutations from routine patient data, potentially enabling faster treatment decisions.

#Interpretable Multi-Instance Learning #Acute Myeloid Leukemia #Flow Cytometry #Medical AI

arXiv cs.LG@Jonathan Legrand et al. (6)Source

Sep 1

Papers73°10:44

Chinese content毫米波无蜂窝大规模MIMO波束成形设计

研究人员用GNN从sub-6 GHz CSI学习毫米波波束成形，比传统方法性能更好，训练开销更低。

#GNN #CFmMIMO #毫米波 #波束成形

arXiv cs.LG@Sina Tavakolian et al. (5)Source

Aug 29

Papers00:35

Chinese content超大规模MIMO系统实现空中学习机分类

康奈尔大学团队推出超大规模MIMO系统，用物理超表面实现机器学习，波域处理精度媲美数字模型。