Mysalashravani·4h agoMachine Learning Workflow & Data PreprocessingA practical guide to the end-to-end ML pipeline, preparing data, training models, evaluating performance, and avoiding data leakage.
Philippe Vandenbroeck·2d agoLearning with Teeth: Evaluation as Democratic PracticeKeynote speech delivered at the bi-annual conference of the European Evaluation Society “Evaluation for Vibrant Democracies”, Lille, 28–30…
Nithin Bharadhwaj·2d agoValid JSON Is Not a Correct Tool Call: A Small Eval Harness for Agents That Change ThingsSchema validation guarantees syntax. Your agent’s real failures live in semantics, retries and side effects — here is a ~100-line harness…
InGoogle Cloud - CommunitybyAbdul Salam·2d agoWhy Your Enterprise AI Agent Passed the Demo and Stalled in UAT: A Field Guide to Golden DatasetsModels and agent frameworks depreciate every quarter. Your ground-truth contract is the only asset that compounds.
InBeingWellbyDr. Patricia Farrell·3d agoHas Autism Become a Label for Everyone?A famous autism scientist says the diagnosis has stretched too far. Autistic physicians say she has it backward.A response icon4A response icon4
Florencia Guerzovich·3d agoKeep the work, Change the Recipe (Post 4)What fidelity means when context changes
Igor Eduardo·3d agoCited Is Not Calibrated: Your Research Agent Sounds More Certain Than Its SourcesBy Igor Eduardo · Austin, TX · https://igoreduardo.com
AIDevBuilds·3d agoYour LLM Judge Agrees With Itself. That Is Not an Eval.An LLM judge can repeat its verdicts and still be wrong. Where human ground truth has to enter the eval loop, and what each entry point cost
InGoogle Cloud - CommunitybyAmir Davidoff·4d agoBuild your own custom code benchmarkBuild the Eval Factory for Your Software Factory: Benchmark Coding Agents on Your Own Repo with repo2bench (Built on Harbor), and Optimize…A response icon1A response icon1
Jatinkumar Parmar·4d agoHow WAS Evals Improve AI ModelsAI models have changed a great deal over the years. Across successive generations, they’ve become better at taking requests of all kinds…
