https://groundy.com/articles/kimi-linear-cuts-kv-cache-75-but-recall-remains-the-binding-constraint/ Groundy en 2026-07-29T07:59:50.092Z Kimi Linear Cuts KV Cache 75% but Recall Remains the Binding Constraint https://groundy.com/articles/why-vendor-model-cards-fail-clinical-ethics-procurement/ Groundy en 2026-07-29T04:26:37.416Z Why Vendor Model Cards Fail Clinical Ethics Procurement https://groundy.com/articles/harness-vs-scaffold-why-claude-code-and-langgraph-are-not-interchangeable/ Groundy en 2026-07-29T03:24:16.193Z Harness vs Scaffold: Why Claude Code and LangGraph Are Not Interchangeable https://groundy.com/articles/fine-tuning-vs-rag-for-internal-apis-starcoder2-constraints/ Groundy en 2026-07-28T21:59:23.041Z Fine-Tuning vs RAG for Internal APIs: StarCoder2 Constraints https://groundy.com/articles/mcp-tool-discovery-moves-from-hardcoded-config-to-runtime-agent-search/ Groundy en 2026-07-28T21:34:51.793Z MCP Tool Discovery Moves From Hardcoded Config to Runtime Agent Search https://groundy.com/articles/calibrated-llm-monitoring-conformal-prediction-with-drift-detection/ Groundy en 2026-07-28T20:33:27.413Z Calibrated LLM Monitoring: Conformal Prediction with Drift Detection https://groundy.com/articles/mellum2-unverified-why-moe-active-parameters-matter-more-than-total-size/ Groundy en 2026-07-28T19:46:10.394Z Mellum2 Unverified: Why MoE Active Parameters Matter More Than Total Size https://groundy.com/articles/code-as-action-agents-beat-gaia-but-require-runtime-sandboxing/ Groundy en 2026-07-28T08:42:18.552Z Code-as-Action Agents Beat GAIA But Require Runtime Sandboxing https://groundy.com/articles/trident-benchmark-llm-safety-gaps-in-finance-medicine-and-law/ Groundy en 2026-07-28T07:44:31.879Z TRIDENT Benchmark: LLM Safety Gaps in Finance, Medicine, and Law https://groundy.com/articles/why-chat-leaderboards-do-not-predict-image-quality/ Groundy en 2026-07-28T06:44:08.852Z Why Chat Leaderboards Do Not Predict Image Quality https://groundy.com/articles/pypi-wheel-reproducibility-15-byte-identical-79-source-equivalent/ Groundy en 2026-07-27T19:43:17.162Z PyPI Wheel Reproducibility: 15% Byte-Identical, 79% Source-Equivalent https://groundy.com/articles/kimi-k3-procurement-governance-review-over-phantom-government-assessments/ Groundy en 2026-07-27T18:37:58.304Z Kimi K3 Procurement: Governance Review Over Phantom Government Assessments https://groundy.com/articles/eu-ai-act-traceability-why-ml-pipelines-fail-conformity-assessment/ Groundy en 2026-07-27T08:07:44.247Z EU AI Act Traceability: Why ML Pipelines Fail Conformity Assessment