{"type":"post","id":"hcltech-researchers-release-iwibench-a-benchmark-for-models-08782f","slug":"hcltech-researchers-release-iwibench-a-benchmark-for-models-08782f","title":"HCLTech researchers release IWIBench, a benchmark for models editing another model's weights","description":"HCLTech researchers published IWIBench, a benchmark testing whether one model can change another model’s behavior by editing its weights.","retrieval_nugget":"HCLTech researchers published IWIBench, a benchmark testing whether one model can change another model’s behavior by editing its weights. Its controlled tasks measure both whether an edit works and whether the target retains other correct behaviors. The results offer a first concrete basis for comparing weight-editing systems, while showing that performance varies sharply across evaluated editors.","published_at":"2026-11-05","updated_at":"2026-11-05","record_date":"2026-11-05","date_kind":"published_at","topics":["models"],"entities":["HCLTech","IWIBench","Gemma 3 4B IT","Gemma 3 1B IT","Chris von Csefalvay","Tamás Foldi","Pranav Doma","Achyuthan Unni Krishnan"],"related_patterns":["ai-engineering-evaluations-are-splitting-into-task-specific"],"novelty_label":"new_fact","what_changed":"HCLTech researchers released IWIBench, a benchmark for testing weight edits across ten task families and whether edited models preserve other behavior; the prior coverage focused on MIT’s weight-space programming work.","prior_refs":[],"source_url":"https://hcltech-robotics.github.io/iwibench","source_urls":["https://hcltech-robotics.github.io/iwibench"],"schema_version":"newruntime-agent-readable-v0.2","stable_id":"post:hcltech-researchers-release-iwibench-a-benchmark-for-models-08782f","status":"published","visuals":[],"editor":{"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]},"routes":{"html":"https://newruntime.com/posts/hcltech-researchers-release-iwibench-a-benchmark-for-models-08782f/","markdown":"https://newruntime.com/posts/hcltech-researchers-release-iwibench-a-benchmark-for-models-08782f.md","json":"https://newruntime.com/posts/hcltech-researchers-release-iwibench-a-benchmark-for-models-08782f.json"},"markdown_url":"https://newruntime.com/posts/hcltech-researchers-release-iwibench-a-benchmark-for-models-08782f.md","next_reads":[{"type":"topic","path":"/topics/models/","url":"https://newruntime.com/topics/models/","title":"Models","reason":"Explore the models topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/models.md","json_url":"https://newruntime.com/topics/models.json"},{"type":"collection","path":"/posts/","url":"https://newruntime.com/posts/","title":"Field Notes","reason":"Open the latest editorial analysis.","media_type":"text/html","json_url":"https://newruntime.com/posts.json"},{"type":"collection","path":"/sources/","url":"https://newruntime.com/sources/","title":"Sources","reason":"See the original sources this archive cites.","media_type":"text/html"}]}