A Shared Subcircuit Lets LLMs Count Down Across Tasks
大模型内部发现通用倒数机制,助你在多种任务中精准控制输出长度
arXiv:2607.12279v1 Announce Type: cross Abstract: Writing a sentence of exactly twelve words; ending a DNA sequence at the right codon; formatting an …
大模型内部发现通用倒数机制,助你在多种任务中精准控制输出长度
arXiv:2607.12279v1 Announce Type: cross Abstract: Writing a sentence of exactly twelve words; ending a DNA sequence at the right codon; formatting an …
多模态大模型视觉计数能力新基准,HoloCount全面评测模型在复杂场景下的计数表现,揭示当前MLLMs的瓶颈。
arXiv:2607.06420v1 Announce Type: new Abstract: Visual counting is a fundamental pillar of multimodal intelligence, requiring a seamless integration o…
一只可爱小猫帮你精准监控Claude和Codex的Token消耗,实时成本一目了然
i, like you, probably burn tokens like water and i don't trust the /status or /usage calls b/c, for whatever reason, they either fail to load in the C…
浏览器内实时运行,基于真实tokenizer的LLM token计数与成本估算工具,无需上传数据。
Article URL: https://holaclaw.ai/tools/token-studio Comments URL: https://news.ycombinator.com/item?id=48491836 Points: 2 # Comments: 1
针对跨域计数中密度组成变化导致模型退化问题,提出条件特征对齐方法,避免全局域不变性带来的负面影响
arXiv:2506.17137v3 Announce Type: replace Abstract: Object counting models often degrade under cross-domain deployment because density composition var…
一分钟搞定AI团队成本分摊,用AgentColony Auditor解析OpenAI使用日志,轻松回答「哪个团队花了多少钱」的FinOps难题。
Request-level AI cost attribution is the fastest way to answer the FinOps question that matters most: which team generated which bill. A usable usage …
无需 API 密钥即可统计 Claude Code 输入 tokens 并估算费用,精准控制开发预算。
Article URL: https://github.com/samjoch/tok Comments URL: https://news.ycombinator.com/item?id=48358986 Points: 3 # Comments: 0
联合国谴责决议统计分析,看哪些国家被点名最多
Article URL: https://boxed.github.io/UN-condemns/ Comments URL: https://news.ycombinator.com/item?id=48338358 Points: 2 # Comments: 0
本地代理实时去重缓存,轻松省下40%-70%的Claude Code账单,API密钥永不泄露。
A local proxy that dedupes repeated context, caches tool results, summarizes long conversations, and streams a live savings counter. Your ANTHROPIC_AP…