1
速腾聚创 2026 上半年激光雷达销量 71.92 万台，机器人业务同比大增 510%
速腾聚创上半年激光雷达销量超71万台，机器人业务暴增510%，自研芯片架构亮相。
IT之家 7 月 9 日消息，速腾聚创今日发布自愿公告，披露了截至 2026 年 6 月 30 日止六个月的未经审核运营数据。 公告显示，公司激光雷达产品上半年总销量达 719,200 台，同比增长 170%。其中，用于 ADAS（IT之家注：高级驾驶辅助系统）应用的激光雷达销量为 436,600 …
2
Open Problem: Is AdamW Effective Under Heavy-Tailed Noise?
AdamW在重尾噪声下是否有效？这篇论文提出了一个亟待解决的开放问题。
arXiv:2606.23676v1 Announce Type: cross Abstract: AdamW is the de facto optimizer for training large language models (LLMs), yet the theory behind it …
3
Show HN: Revenant – automatic LLM powered reverse engineering
LLM驱动的全自动逆向工程工具，集成radare2和ghidra，一键分析固件并生成开源框架，高效解放硬件安全研究员。
I am a hardware engineer and security researcher and I've been wondering whether my work could be partially automated, so I can focus on other topics …
4
Adafruit files suit against Flux.ai over legal threats [pdf]
开源硬件巨头Adafruit起诉AI电路设计平台Flux.ai，指控其不正当竞争与法律威胁，揭示开源社区与商业工具的冲突。
Article URL: https://storage.courtlistener.com/recap/gov.uscourts.cand.471648/gov.uscourts.cand.471648.1.0.pdf Comments URL: https://news.ycombinator.…
5
AdaGC: Enhancing LLM Pretraining Stability via Adaptive Gradient Clipping
被ICML接收的自适应梯度裁剪方法，有效提升LLM预训练稳定性，AI训练优化的新突破
arXiv:2502.11034v3 Announce Type: replace Abstract: Loss spikes remain a persistent obstacle in large-scale language model pretraining. While previous…
6
Muon Learns More Robust and Transferable Features than Adam
Muon优化器比Adam学习到更鲁棒和可迁移的特征，为深度学习训练提供新选择。
arXiv:2606.09658v1 Announce Type: new Abstract: Muon has recently emerged as a state-of-the-art optimizer for pretraining Large Language Models (LLMs)…
7
Cross-Epoch Adaptive Rollout Optimization for RL Post-Training
面向RL后训练的自适应跨epoch rollout优化新方法，提升强化学习训练效率。
arXiv:2606.05606v1 Announce Type: cross Abstract: LLM post-training often relies on reinforcement learning methods that sample multiple rollouts per p…
8
Adaptive Latent Agentic Reasoning
自适应潜空间推理，让LLM代理告别冗长低效的思维链，大幅提升推理效率。
arXiv:2606.02871v1 Announce Type: cross Abstract: Large reasoning models improve performance by generating extended chain-of-thought (CoT) reasoning, …
9
Training Prompt Matters: State-Adaptive Optimization for Robust Fine-Tuning
提出状态自适应优化方法，揭示训练提示词对模型鲁棒微调的关键影响。
arXiv:2606.01967v1 Announce Type: new Abstract: While prompt engineering is instrumental in maximizing the capabilities of Large Language Models (LLMs…
10
Show HN: Glq LLM quantization using E8 lattice
基于E8晶格的新型LLM量化方法GLQ，8权重一组映射为16位索引，精度2-8 bpw，开源有代码。
I have with the help of AI create an open source method of E8 LLM code book quantization library called glq. I was interested in creating Glq as a PC …
11
Adafruit Receives Demand Letter from Fenwick Legal Counsel on Behalf of Flux.ai
开源硬件巨头遭Flux.ai律师函，法律博弈揭示开源与商业化的敏感边界
Article URL: https://blog.adafruit.com/ Comments URL: https://news.ycombinator.com/item?id=48368121 Points: 262 # Comments: 91
12
AdaSD: Adaptive Speculative Decoding for Efficient Language Model Inference
自适应投机解码新方法AdaSD，大幅提升大模型推理效率，值得关注的最新论文成果
arXiv:2512.11280v2 Announce Type: replace Abstract: Large language models (LLMs) have achieved remarkable performance across a wide range of tasks, bu…
13
Revisiting the Adam-SGD Gap in LLM Pre-Training: The Role of Large Effective Learning Rates
揭秘SGD在LLM预训练中不如Adam的根源：大有效学习率的关键作用。
arXiv:2605.17787v1 Announce Type: new Abstract: It is widely believed that stochastic gradient descent (SGD) performs significantly worse than adaptiv…
14
Prompt Compression in Diffusion Large Language Models: Evaluating LLMLingua-2 on LLaDA
评估LLMLingua-2在扩散大模型LLaDA上的提示压缩效果，探索高效推理新路径。
arXiv:2605.17932v1 Announce Type: new Abstract: Prompt compression reduces inference cost and context length in large language models, but prior evalu…
15
No-signup link unfurl for AI agents (an agent can't do a signup)
无需注册、无需API密钥，一条GET请求即可获取链接元数据，专为无法完成注册的AI智能体设计。
If you write code that calls links — a bot, a crawler, a RAG ingest job, an autonomous agent — you eventually need the metadata behind a URL: title, d…
16
Using GPT-4 to deliver a new customer service standard
Ada借助GPT-4打造全新客服标准，效率与质量双提升
Ada uses GPT-4 to deliver a new customer service standard
17
Building efficient MCP servers
MCP协议正从低效的SSE向Streamable HTTP演进，Vercel的MCP adapter帮你快速构建高效、可扩展的MCP服务器，适配主流框架，让AI集成不掉队。
The Model Context Protocol (MCP) standardizes how to build integrations for AI models. We built the MCP adapter to help developers create their own MC…