{"type":"post","id":"qwenpu-compiles-an-npu-into-qwen3-5-4b-077c19","slug":"qwenpu-compiles-an-npu-into-qwen3-5-4b-077c19","title":"QwenPU compiles an NPU into Qwen3.5-4B's weights and runs Karpathy's stories260K language model inside it","description":"QwenPU is a research project that embeds a vector-processing unit in a modified Qwen3.5-4B checkpoint, then runs Karpathy’s 260,000-parameter stories260K model inside that unit.","retrieval_nugget":"QwenPU is a research project that embeds a vector-processing unit in a modified Qwen3.5-4B checkpoint, then runs Karpathy’s 260,000-parameter stories260K model inside that unit. Its significance is a demonstration of exact, inspectable computation within a model’s forward pass—not a ready-to-use way to speed up inference.","published_at":"2026-10-31","updated_at":"2026-10-31","record_date":"2026-10-31","date_kind":"published_at","topics":["harness-engineering","agent-runtime"],"entities":["QwenPU","Qwen3.5-4B","Qwenputer","Divyansh Agrawal","stories260K","Andrej Karpathy"],"related_patterns":["harness-architecture-outlives-model-choice"],"novelty_label":"new_fact","what_changed":"QwenPU describes compiling a vector unit into Qwen3.5-4B’s weights so it can run Karpathy’s stories260K; the prior covers a separate Coral NPU architecture.","prior_refs":["memory:92a5f68d-93ae-8e05-aaba-366e5a47002e"],"source_url":"https://github.com/divagr18/QwenPU","source_urls":["https://github.com/divagr18/QwenPU"],"schema_version":"newruntime-agent-readable-v0.2","stable_id":"post:qwenpu-compiles-an-npu-into-qwen3-5-4b-077c19","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/qwenpu-compiles-an-npu-into-qwen3-5-4b-077c19/","markdown":"https://newruntime.com/posts/qwenpu-compiles-an-npu-into-qwen3-5-4b-077c19.md","json":"https://newruntime.com/posts/qwenpu-compiles-an-npu-into-qwen3-5-4b-077c19.json"},"next_reads":[{"type":"topic","path":"/topics/harness-engineering/","url":"https://newruntime.com/topics/harness-engineering/","title":"Harness Engineering","reason":"Explore the harness-engineering topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/harness-engineering.md","json_url":"https://newruntime.com/topics/harness-engineering.json"},{"type":"topic","path":"/topics/agent-runtime/","url":"https://newruntime.com/topics/agent-runtime/","title":"Agent Runtime","reason":"Explore the agent-runtime topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/agent-runtime.md","json_url":"https://newruntime.com/topics/agent-runtime.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":"/source-ledger.json","url":"https://newruntime.com/source-ledger.json","title":"Source Ledger","reason":"Inspect the public source evidence graph.","media_type":"application/json"}]}