📄 文章列表

840 篇文章
TDS

Your LLM Has a Curved Space of Paragraphs

✅ completed ⭐ ☆☆☆☆☆ 2026-09-26
VentureBeat

AI agents have routed around access blocks. Only 9% of companies in VentureBeat's August survey isolate high-risk agents

✅ completed ⭐ ☆☆☆☆☆ 2026-09-25
TDS

Beyond RAGs: Building Actually Truthful AI Harnesses

✅ completed ⭐ ☆☆☆☆☆ 2026-09-25
VentureBeat

AI coding tools are accelerating dependency sprawl and expanding malware risk with it

✅ completed ⭐ ☆☆☆☆☆ 2026-09-24
TDS

How GRPO Trains Small Language Models with Verifiable Rewards

✅ completed ⭐ ☆☆☆☆☆ 2026-09-24
TDS

When the Correct Answer Is Nothing, What Does Your Pipeline Return?

✅ completed ⭐ ☆☆☆☆☆ 2026-09-24
TDS

Towards Spec-Driven Test Automation: Part 1

✅ completed ⭐ ☆☆☆☆☆ 2026-09-24
VentureBeat

Anthropic releases Claude Opus 5.5, beating Fable 5.1 on key agentic benchmarks at 60% cheaper API price

✅ completed ⭐ ☆☆☆☆☆ 2026-09-22
VentureBeat

Meta patched Muse’s zero-day, but security teams still lack visibility into what the agent can access

✅ completed ⭐ ☆☆☆☆☆ 2026-09-22
VentureBeat

'Better than DeepSeek': Xiaomi's MiMo-V2.6-Pro debuts as the top open weights model in the world alongside cheaper V2.6-

✅ completed ⭐ ☆☆☆☆☆ 2026-09-22
VentureBeat

Companies are putting Jev in charge of AI agent decisions — and prompt injection can influence the verdict

✅ completed ⭐ ☆☆☆☆☆ 2026-09-21
TDS

Your AI Assistant Wrote the Code. Who Checked the Defaults?

✅ completed ⭐ ☆☆☆☆☆ 2026-09-21
VentureBeat

Google’s open source EnvHarness lets AI agents train against environments that evolve with them

✅ completed ⭐ ☆☆☆☆☆ 2026-09-21
TDS

One Vendor, Four Spellings: How Deterministic Stages Beat Similarity Scores

✅ completed ⭐ ☆☆☆☆☆ 2026-09-20
TDS

CBAM Paper Walkthrough: The Double-Attention Mechanism

✅ completed ⭐ ☆☆☆☆☆ 2026-09-20
VentureBeat

Inference chips are locked in years before the models they’ll run. Architect Labs says AI can design them in weeks

✅ completed ⭐ ☆☆☆☆☆ 2026-09-18
VentureBeat

69% of enterprises that install OpenAI's agent platform make it primary. For Anthropic's Claude Platform it's 38%

✅ completed ⭐ ☆☆☆☆☆ 2026-09-19
VentureBeat

An AI system designed an inference chip in two weeks. Now Architect Labs has to prove it in silicon

✅ completed ⭐ ☆☆☆☆☆ 2026-09-19
TDS

Building a Data Lakehouse with DuckDB and DuckLake

✅ completed ⭐ ☆☆☆☆☆ 2026-09-18
TDS

Multi-Agent Coding Isn’t Enough — Agents Need a Commitment Layer

✅ completed ⭐ ☆☆☆☆☆ 2026-09-18