AIトレンド(βテスト)

新興トピック発見フィード

最終更新: 2026-07-06(UTC基準)/集計窓: 直近約7日

本日のテーマ抽出(AI生成)は未実施のため、直近約7日の「生シグナル」をそのまま表示しています。各項目は一次ソースへのリンク付きです。

スター急増リポジトリ

リポジトリ勢い
open-webui/open-webuiスター増分 +67(現在 144,339)
Comfy-Org/ComfyUIスター増分 +62(現在 119,551)
ggml-org/llama.cppスター増分 +52(現在 119,380)
langgenius/difyスター増分 +48(現在 147,801)
langchain-ai/langchainスター増分 +43(現在 141,018)
openai/whisperスター増分 +42(現在 104,266)
ollama/ollamaスター増分 +40(現在 175,549)
langchain-ai/langgraphスター増分 +36(現在 36,563)
vllm-project/vllmスター増分 +30(現在 85,441)
crewAIInc/crewAIスター増分 +27(現在 54,968)
ggml-org/whisper.cppスター増分 +21(現在 51,320)
huggingface/transformersスター増分 +20(現在 162,284)
modelcontextprotocol/serversスター増分 +16(現在 88,093)
lobehub/lobehubスター増分 +16(現在 79,494)
run-llama/llama_indexスター増分 +8(現在 50,664)
microsoft/autogenスター増分 +7(現在 59,511)
chroma-core/chromaスター増分 +7(現在 28,699)
modelcontextprotocol/typescript-sdkスター増分 +6(現在 12,801)
AUTOMATIC1111/stable-diffusion-webuiスター増分 +3(現在 163,997)
modelcontextprotocol/python-sdkスター増分 +3(現在 23,537)

新着 arXiv 論文(cs.AI / cs.CL / cs.LG)

タイトルカテゴリ公開日
Distributed Attacks in Persistent-State AI Controlcs.AI2026-07-02
LACUNA: A Testbed for Evaluating Localization Precision for LLM Unlearningcs.CL, cs.AI, cs.LG2026-07-02
Program-as-Weights: A Programming Paradigm for Fuzzy Functionscs.LG, cs.AI, cs.CL2026-07-02
Online Safety Monitoring for LLMscs.AI, cs.CL, cs.LG, stat.AP, stat.ML2026-07-02
ReContext: Recursive Evidence Replay as LLM Harness for Long-Context Reasoningcs.AI2026-07-02
What LLM Agents Say When No One Is Watching: Social Structure and Latent Objective Emergence in Multi-Agent Debatescs.AI, cs.CL, cs.LG, cs.MA2026-07-02
Reasoning LLM Improves Speaker Recognition in Long-form TV Dramascs.CL, cs.AI, cs.CV2026-07-02
DemoPSD: Disagreement-Modulated Policy Self-Distillationcs.LG, cs.AI2026-07-02
Beyond Adam: SOAP and Muon for Faster, Label-Efficient Training of Machine Learning Interatomic Potentialscs.LG, cs.AI, physics.chem-ph, physics.comp-ph2026-07-02
Controllable Sim Agents with Behavior Latentscs.RO, cs.LG2026-07-02
Towards Robustness against Typographic Attack with Training-free Concept Localizationcs.CV, cs.CL2026-07-02
G-RRM: Guiding Symbolic Solvers with Recurrent Reasoning Modelscs.AI2026-07-02
Visually Grounded Self-Reflection for Vision-Language Models via Reinforcement Learningcs.CL, cs.CV2026-07-02
Combating Textual Noise and Redundancy: Entropy-Aware Dense Visual Token Pruningcs.CV, cs.AI2026-07-02
Audio-Based Understanding of Audiobook Narration Appealcs.CL, cs.SD, eess.AS2026-07-02
TestEvo-Bench: An Executable and Live Benchmark for Test and Code Co-Evolutioncs.SE, cs.AI, cs.CL2026-07-02
Human Capital, Not Model Benchmarks, Predicts Hybrid Intelligence in Forecastingcs.CY, cs.AI2026-07-02
Learning to Move Before Learning to Do: Task-Agnostic pretraining for VLAscs.RO, cs.AI2026-07-02
Will Scaling Improve Social Simulation with LLMs?cs.CL2026-07-02
OrbitQuant: Data-Agnostic Quantization for Image and Video Diffusion Transformerscs.CV, cs.AI, cs.LG2026-07-02

Hacker News 上位スレッド

タイトルPoints
Jamesob's guide to running SOTA LLMs locally [HN]405
Show HN: Bramble – Local-first password manager [HN]142
Launch HN: Manufact (YC S25) – MCP Cloud [HN]108
Show HN: KiCad in the Browser [HN]95
Egg consumption inversely correlated with Alzheimer's [HN]66
Mouse: Precision Editing Tools for AI Coding Agents [HN]37
The Plight of the Martian Farmer [HN]35
Efficacy and Safety of Psilocybin in Treatment-Resistant Major Depression [HN]34
Reducing Assumptions, Exploding Your Code [HN]24
Does Code Cleanliness Affect Coding Agents? [HN]23
Anthropic performing prompt injection on its users [HN]21
Connections in Math: the two kinds of random [HN]20
Mapping with In-Memory Layers to Reduce LLM Overload [HN]16
Tell HN: don't trust Bigco AI agents with AI research IP [HN]16
Backon – Python retry (zero deps, circuit breaker, async native) [HN]13
Show HN: Enola-A deterministic architecture graph for developers and AI agents [HN]10
Context graphs: how AI agents can store and use past decisions [HN]9
Portugal launches first open-source AI model, joining Europe's sovereignty push [HN]8
Building Agents That Don't Break Themselves [HN]8
Damo Academy unveils an AI agent able to discover superconductors [HN]7

Hugging Face DL数上位

モデル/データ種別DL数
sentence-transformers/all-MiniLM-L6-v2model244,099,047
cross-encoder/ms-marco-MiniLM-L6-v2model79,982,687
google-bert/bert-base-uncasedmodel62,774,318
BAAI/bge-small-en-v1.5model61,557,234
sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2model47,685,110
google/electra-base-discriminatormodel39,011,105
sentence-transformers/all-mpnet-base-v2model33,023,896
BAAI/bge-m3model32,338,205
Qwen/Qwen3-0.6Bmodel27,835,708
timm/mobilenetv3_small_100.lamb_in1kmodel26,534,440
openai/clip-vit-base-patch32model22,262,188
FacebookAI/xlm-roberta-basemodel20,506,108
google-t5/t5-smallmodel19,836,097
Qwen/Qwen3-8Bmodel16,401,211
nomic-ai/nomic-embed-text-v1.5model16,250,819
BAAI/bge-reranker-v2-m3model16,223,993
amazon/chronos-2model15,157,645
BAAI/bge-large-en-v1.5model14,072,512
facebook/opt-125mmodel13,869,888
hexgrad/Kokoro-82Mmodel13,685,663