{"version":"1.0","generated_at":"2026-10-04T22:16:14.249416","id":1750,"slug":"deepseek-v4-flash-amd-instinct-mi355x-gpu-miles","title":"DeepSeek-V4 Flash 强化学习训练登陆 AMD Instinct MI355X GPU，由 Miles 框架支持","summary":"DeepSeek-V4 Flash 的强化学习训练已在 AMD Instinct MI355X GPU 上通过 Miles 框架成功运行，解决了三大技术挑战，验证了性能提升。","abstract":"DeepSeek-V4 Flash 强化学习训练登陆 AMD Instinct MI355X GPU，由 Miles 框架支持 DeepSeek-V4 Flash 的强化学习训练已在 AMD Instinct MI355X GPU 上通过 Miles 框架成功运行，解决了三大技术挑战，验证了性能提升。 DeepSeek-V4 Flash 的强化学习训练已在 AMD Instinct MI355X GPU 上通过 Miles 框架成功运行，解决了三大技术挑战，验证了性能提升。 原贴提到：DeepSeek-V4 Flash 的强化学习训练现已在 AMD Instinct MI355X GPU 上通过 Miles 框架获得支持，基于 来源：lmsys.org DeepSeek-V4 Flash 的强化学习训练现已在 AMD Instinct MI355X GPU 上通过 Miles 框架获得支持，基于 ROCm 软件栈运行。该 2840 亿参数 MoE 模型（每 token 激活 130 亿参数）需 SGLang 进行 rollout 生成、Megatron 进行策略更新，Miles 负责异步循环与权重同步。团队解决了 SGLang 与 Megatron 间模型对齐、量化状态在线更新及多节点并行稳定性三大挑战，最终在四个八 GPU 节点上完成端到端验证：超过 100 个优化器步骤中训练-rollout 对数概率差可控，在线奖励持续提升，离线 AIME-2024 基准分数同步上涨。 这是什么信号？ DeepSeek 正在将强化学习训练迁移到 AMD 硬件平台，这意味着 AI 大模型训…","access_level":"member","access_label":"免费阅读","access_mode":"full","is_preview":false,"canonical_url":"https://opc.beizhux.com/content/1750/deepseek-v4-flash-amd-instinct-mi355x-gpu-miles","html_url":"https://opc.beizhux.com/content/1750/deepseek-v4-flash-amd-instinct-mi355x-gpu-miles","json_url":"https://opc.beizhux.com/content/1750/deepseek-v4-flash-amd-instinct-mi355x-gpu-miles.json","published_at":"2026-07-11T01:19:11","updated_at":"2026-10-04T20:55:31","category":{"slug":"ai-business","name":"AI小生意项目库"},"source":{"site":"lmsys.org","author":"LMSYS：Blog（Chatbot Arena 团队）","url":"https://www.lmsys.org/blog/2026-07-10-rocm-miles-dsv4"},"tags":["AI","AIHOT 精选","AI训练","AMD","AMD GPU","AMD Instinct","AMD Instinct MI355X","Deepseek","DeepSeek-V4","DeepSeek-V4 Flash","GPU","GPU训练"],"topics":[{"slug":"ai-side-business","name":"AI副业","url":"https://opc.beizhux.com/topics/ai-side-business","reason":"这篇内容命中「项目、小生意、变现」等主题信号。"},{"slug":"ai-tools","name":"AI工具","url":"https://opc.beizhux.com/topics/ai-tools","reason":"这篇内容命中「模型」等主题信号。"},{"slug":"ai-super-individual","name":"AI超级个体","url":"https://opc.beizhux.com/topics/ai-super-individual","reason":"这篇内容命中「学习」等主题信号。"}],"keywords":["AI小生意项目库","AI副业","AI工具","AI超级个体","工具","自动化","模型","Cursor","Claude","工作流","副业","创业"],"questions":[{"question":"DeepSeek-V4 Flash 强化学习训练登陆 AMD Instinct MI355X GPU，由 Miles 框架支持主要讲什么？","answer":"DeepSeek-V4 Flash 的强化学习训练已在 AMD Instinct MI355X GPU 上通过 Miles 框架成功运行，解决了三大技术挑战，验证了性能提升。"},{"question":"这篇文章最值得关注的要点是什么？","answer":"DeepSeek-V4 Flash 的强化学习训练已在 AMD Instinct MI355X GPU 上通过 Miles 框架成功运行，解决了三大技术挑战，验证了性能提升。；原贴提到：DeepSeek-V4 Flash 的强化学习训练现已在 AMD Instinct MI355X GPU 上通过 Miles 框架获得支持，基于；来源：lmsys.org"},{"question":"这篇文章和哪些AI专题相关？","answer":"它适合放在AI副业、AI工具、AI超级个体专题里阅读。 关联原因：这篇内容命中「项目、小生意、变现」等主题信号。；这篇内容命中「模型」等主题信号。；这篇内容命中「学习」等主题信号。"},{"question":"阅读这篇文章建议先理解哪些关键词？","answer":"建议先理解AI工具、工具、自动化、模型、Cursor这些关键词，再结合正文判断工具、机会或风险是否值得进入自己的工作流。"}],"terms":[{"slug":"ai-tools-term","name":"AI工具","definition":"在AI觉醒星球里，「AI工具」属于「AI工具」方向。围绕AI工具、模型能力、自动化流程和真实使用场景做整理，优先关注能提升效率、降低成本和创造新交付的工具。 不只看工具热度，更看它能不能进入真实流程。","topic_slug":"ai-tools","topic_name":"AI工具","topic_title":"AI工具：模型、插件、自动化与实战场景","topic_url":"https://opc.beizhux.com/topics/ai-tools","topic_path":"/topics/ai-tools","url":"https://opc.beizhux.com/glossary/ai-tools-term","path":"/glossary/ai-tools-term","json_url":"https://opc.beizhux.com/glossary/ai-tools-term.json"},{"slug":"tools","name":"工具","definition":"在AI觉醒星球里，「工具」属于「AI工具」方向。围绕AI工具、模型能力、自动化流程和真实使用场景做整理，优先关注能提升效率、降低成本和创造新交付的工具。 不只看工具热度，更看它能不能进入真实流程。","topic_slug":"ai-tools","topic_name":"AI工具","topic_title":"AI工具：模型、插件、自动化与实战场景","topic_url":"https://opc.beizhux.com/topics/ai-tools","topic_path":"/topics/ai-tools","url":"https://opc.beizhux.com/glossary/tools","path":"/glossary/tools","json_url":"https://opc.beizhux.com/glossary/tools.json"},{"slug":"automation","name":"自动化","definition":"在AI觉醒星球里，「自动化」属于「AI工具」方向。围绕AI工具、模型能力、自动化流程和真实使用场景做整理，优先关注能提升效率、降低成本和创造新交付的工具。 不只看工具热度，更看它能不能进入真实流程。","topic_slug":"ai-tools","topic_name":"AI工具","topic_title":"AI工具：模型、插件、自动化与实战场景","topic_url":"https://opc.beizhux.com/topics/ai-tools","topic_path":"/topics/ai-tools","url":"https://opc.beizhux.com/glossary/automation","path":"/glossary/automation","json_url":"https://opc.beizhux.com/glossary/automation.json"},{"slug":"ai-model","name":"模型","definition":"在AI觉醒星球里，「模型」属于「AI工具」方向。围绕AI工具、模型能力、自动化流程和真实使用场景做整理，优先关注能提升效率、降低成本和创造新交付的工具。 不只看工具热度，更看它能不能进入真实流程。","topic_slug":"ai-tools","topic_name":"AI工具","topic_title":"AI工具：模型、插件、自动化与实战场景","topic_url":"https://opc.beizhux.com/topics/ai-tools","topic_path":"/topics/ai-tools","url":"https://opc.beizhux.com/glossary/ai-model","path":"/glossary/ai-model","json_url":"https://opc.beizhux.com/glossary/ai-model.json"},{"slug":"cursor","name":"Cursor","definition":"在AI觉醒星球里，「Cursor」属于「AI工具」方向。围绕AI工具、模型能力、自动化流程和真实使用场景做整理，优先关注能提升效率、降低成本和创造新交付的工具。 不只看工具热度，更看它能不能进入真实流程。","topic_slug":"ai-tools","topic_name":"AI工具","topic_title":"AI工具：模型、插件、自动化与实战场景","topic_url":"https://opc.beizhux.com/topics/ai-tools","topic_path":"/topics/ai-tools","url":"https://opc.beizhux.com/glossary/cursor","path":"/glossary/cursor","json_url":"https://opc.beizhux.com/glossary/cursor.json"},{"slug":"claude","name":"Claude","definition":"在AI觉醒星球里，「Claude」属于「AI工具」方向。围绕AI工具、模型能力、自动化流程和真实使用场景做整理，优先关注能提升效率、降低成本和创造新交付的工具。 不只看工具热度，更看它能不能进入真实流程。","topic_slug":"ai-tools","topic_name":"AI工具","topic_title":"AI工具：模型、插件、自动化与实战场景","topic_url":"https://opc.beizhux.com/topics/ai-tools","topic_path":"/topics/ai-tools","url":"https://opc.beizhux.com/glossary/claude","path":"/glossary/claude","json_url":"https://opc.beizhux.com/glossary/claude.json"},{"slug":"workflow","name":"工作流","definition":"在AI觉醒星球里，「工作流」属于「Agent工作流」方向。聚合Agent、AI编程、Claude Code、Cursor、工作流自动化等内容，帮助你把智能体能力落到真实任务链路。 把智能体当成任务系统，而不是聊天窗口。","topic_slug":"agent-workflow","topic_name":"Agent工作流","topic_title":"Agent工作流：智能体、AI编程与自动化协作","topic_url":"https://opc.beizhux.com/topics/agent-workflow","topic_path":"/topics/agent-workflow","url":"https://opc.beizhux.com/glossary/workflow","path":"/glossary/workflow","json_url":"https://opc.beizhux.com/glossary/workflow.json"},{"slug":"side-business","name":"副业","definition":"在AI觉醒星球里，「副业」属于「AI副业」方向。筛选AI副业、AI小生意、项目认知和可收费交付案例，关注普通人可以验证、复制和沉淀的方法。 从真实痛点出发，先做能收费的小验证。","topic_slug":"ai-side-business","topic_name":"AI副业","topic_title":"AI副业：小生意、项目机会与可收费交付","topic_url":"https://opc.beizhux.com/topics/ai-side-business","topic_path":"/topics/ai-side-business","url":"https://opc.beizhux.com/glossary/side-business","path":"/glossary/side-business","json_url":"https://opc.beizhux.com/glossary/side-business.json"}],"preview_hidden_sections":[],"related_items":[{"id":3223,"title":"美财长贝森特批评 Anthropic、OpenAI 等风险警告：危言耸听、光喊话不拿解决方案","url":"https://opc.beizhux.com/content/3223/anthropic-openai","summary":"美国财政部长贝森特在 Axios 采访中点名批评 Anthropic CEO 阿莫代伊等人的 AI 生存性风险警告，认为只发出危言耸听的声音却拿不出解决方案不是领导力，并呼吁实验室负责人承担安全责任、安全地加速发展。","access_level":"member"},{"id":3224,"title":"BestBlogs 早报：VS Code 周更与递归语言模型等十则","url":"https://opc.beizhux.com/content/3224/bestblogs-vs-code","summary":"BestBlogs 早报汇总十则 AI 动态，其中三条值得单独拎出：VS Code 团队把发布节奏从月更改为周更，代码存活率由 55% 升至 86%；Google DeepMind 的 Interactions API 用交互标识保存上下文与思维签名，并以环境标识复用托管智能体沙箱；DatologyAI 把预训练合成…","access_level":"member"},{"id":3226,"title":"AI 用 5 天生成《辐射：纽约》浏览器游戏，被贝塞斯达发函叫停","url":"https://opc.beizhux.com/content/3226/ai-5-2","summary":"创作者 Chrisfirst 借助 Claude Opus 5.5 在五天内生成浏览器游戏《辐射：纽约》，因贝塞斯达发停止侵权函被叫停。该事件凸显 AI 降低游戏开发门槛与 IP 合规风险之间的冲突，可能影响头部游戏 IP 的玩家生态。","access_level":"member"}],"usage_policy":{"summarizable":true,"preferred_url":"https://opc.beizhux.com/content/1750/deepseek-v4-flash-amd-instinct-mi355x-gpu-miles","private_fields_excluded":true}}