{ "version": 1, "event_id": "evt_b083109812b31e54", "url": "https://xiyu.news/events/evt_b083109812b31e54/", "json": "https://xiyu.news/api/events/evt_b083109812b31e54.json", "type": "other", "status": "monitoring", "category": "technology", "title": { "zh": "Qwenputer：Qwen3.5内部一个可运行、确定性的RISC-V CPU", "en": "Qwenputer: A functioning, deterministic RISC-V CPU inside Qwen3.5" }, "current_state": { "zh": "Qwenputer已在Qwen3.5-4B权重内实现可运行的32位RISC-V虚拟CPU和可精确读写内存，能执行编译后的C程序并通过RISC-V指令测试，大数乘法测试得分39/40；但模型体积增至18.8GB且计算速度远不如外部工具。", "en": "Qwenputer has implemented a working 32-bit RISC-V virtual CPU and exact read/write memory inside Qwen3.5-4B weights, can execute compiled C programs and pass RISC-V instruction tests, and scores 39/40 on large-number multiplication; however, the model grows to 18.8GB and is much slower than external tools." }, "first_seen_at": "2026-10-29T06:51:30.535162+00:00", "last_updated_at": "2026-10-29T11:01:16.505757+00:00", "last_material_change_at": "2026-10-29T11:01:16.505757+00:00", "confidence": 0.75, "updates_count": 2, "sources_count": 2, "entities": [ "cpu", "qwen3", "qwenputer", "risc", "risc-v" ], "identifiers": [], "topics": [ "ai-engineering", "deterministic-computation", "gated-delta-networks", "llm", "model-architecture", "qwen", "research", "risc-v" ], "updates": [ { "update_id": "upd_5713433ba65679cb", "event_id": "evt_b083109812b31e54", "occurred_at": "2026-10-29T00:25:45Z", "published_at": "2026-10-29T00:25:45Z", "first_seen_at": "2026-10-29T06:51:30.535162Z", "time_precision": "published", "update_type": "initial", "material_change": true, "title_zh": "Qwenputer：Qwen3.5内部一个可运行、确定性的RISC-V CPU", "title_en": "Qwenputer: A functioning, deterministic RISC-V CPU inside Qwen3.5", "what_changed_zh": "一名开发者利用门控增量网络层作为精确的类RAM存储，在Qwen3.5内部实现了一个可运行、确定性的RISC-V CPU和文件系统。", "what_changed_en": "A developer implemented a functioning, deterministic RISC-V CPU and filesystem inside Qwen3.5 by exploiting gated delta network layers as exact RAM-like storage.", "current_state_zh": "一名开发者利用门控增量网络层作为精确的类RAM存储，在Qwen3.5内部实现了一个可运行、确定性的RISC-V CPU和文件系统。", "current_state_en": "A developer implemented a functioning, deterministic RISC-V CPU and filesystem inside Qwen3.5 by exploiting gated delta network layers as exact RAM-like storage.", "detailed_summary_zh": "A developer implemented a functioning, deterministic RISC-V CPU and filesystem inside Qwen3.5 by exploiting gated delta network layers as exact RAM-like storage.", "detailed_summary_en": "A developer implemented a functioning, deterministic RISC-V CPU and filesystem inside Qwen3.5 by exploiting gated delta network layers as exact RAM-like storage.", "background_zh": "", "background_en": "", "community_discussion_zh": "", "community_discussion_en": "", "market_impact_zh": "", "market_impact_en": "", "importance_score": 7.0, "references": [ { "url": "https://news.ycombinator.com/item?id=50282838", "title": "Community discussion" } ], "confidence": 0.75, "story_ids": [ "hackernews:story:50282838" ], "sources": [ { "url": "https://divagr.com/blog/qwenputer", "label": "csmantle", "source_type": "hackernews", "official": false } ] }, { "update_id": "upd_99fac4b99ac883e8", "event_id": "evt_b083109812b31e54", "occurred_at": "2026-10-29T08:40:01Z", "published_at": "2026-10-29T08:40:01Z", "first_seen_at": "2026-10-29T11:01:16.505757Z", "time_precision": "published", "update_type": "confirmation", "material_change": true, "title_zh": "开发者把32位CPU编进千问权重：大模型内部直接运行计算机程序", "title_en": "开发者把32位CPU编进千问权重：大模型内部直接运行计算机程序", "what_changed_zh": "报道补充了Qwenputer的实现细节和测试结果：开发者为Divyansh Agrawal，在Qwen3.5-4B原有32层前增加11层，将32位RISC-V虚拟CPU编码进权重，并利用Gated DeltaNet实现可精确读写内存；该CPU可运行编译后的C程序，累计执行约1800万条指令未发现错误，在DeepMind 40道大数乘法题中Qwenputer答对39道，而原版Qwen答对22道。方案效率较低，模型文件因大量零值参数增至18.8GB，约为原版两倍。", "what_changed_en": "The report adds implementation details and test results for Qwenputer: developer Divyansh Agrawal prepended 11 layers to Qwen3.5-4B's original 32 layers to encode a 32-bit RISC-V virtual CPU into the weights, using Gated DeltaNet for exact read/write memory; the CPU runs compiled C programs, executed about 18 million instructions without detected errors, and scored 39/40 on DeepMind's large-number multiplication test versus 22/40 for base Qwen. The approach is inefficient, with the model file growing to 18.8GB, about twice the original, due to many zero-valued parameters.", "current_state_zh": "Qwenputer已在Qwen3.5-4B权重内实现可运行的32位RISC-V虚拟CPU和可精确读写内存，能执行编译后的C程序并通过RISC-V指令测试，大数乘法测试得分39/40；但模型体积增至18.8GB且计算速度远不如外部工具。", "current_state_en": "Qwenputer has implemented a working 32-bit RISC-V virtual CPU and exact read/write memory inside Qwen3.5-4B weights, can execute compiled C programs and pass RISC-V instruction tests, and scores 39/40 on large-number multiplication; however, the model grows to 18.8GB and is much slower than external tools.", "detailed_summary_zh": "Independent developer Divyansh Agrawal built Qwenputer by prepending 11 layers to Qwen3.5-4B that encode a 32-bit RISC-V virtual CPU directly into model weights, letting the LLM execute compiled C programs and exact arithmetic internally and scoring 39/40 on DeepMind's large-number multiplication test versus 22/40 for base Qwen.", "detailed_summary_en": "Independent developer Divyansh Agrawal built Qwenputer by prepending 11 layers to Qwen3.5-4B that encode a 32-bit RISC-V virtual CPU directly into model weights, letting the LLM execute compiled C programs and exact arithmetic internally and scoring 39/40 on DeepMind's large-number multiplication test versus 22/40 for base Qwen.", "background_zh": "", "background_en": "", "community_discussion_zh": "", "community_discussion_en": "", "market_impact_zh": "", "market_impact_en": "", "importance_score": 7.0, "references": [], "confidence": 0.95, "story_ids": [ "telegram:theblockbeats:202974" ], "sources": [ { "url": "https://m.theblockbeats.info/flash/374349?from=telegram", "label": "theblockbeats", "source_type": "telegram", "official": false } ] } ] }