IBM企业级AI开发平台watsonx.ai线DeepSeek R1蒸馏模型

IT之家
11 Feb

IT之家 2 月 11 日消息,IBM 日前宣布,DeepSeek-R1 蒸馏版 Llama 3.1 8B 和 Llama 3.3 70B 现已上线 IBM 的企业级 AI 开发平台 watsonx.ai。

根据官方介绍,DeepSeek 还借助知识蒸馏技术,利用 R1 模型生成的数据优化了多个 Llama 和 Qwen 变体。用户可通过以下方式在 watsonx.ai 上使用 DeepSeek 蒸馏模型:

在 watsonx.ai 的“按需部署”目录中,IBM 提供 Llama 蒸馏版,用户可以部署专用实例以进行安全推理。

用户还可通过“自定义基础模型”导入功能,上传 DeepSeek-R1 其他变体,例如 Qwen 蒸馏模型。

DeepSeek-R1 具备强大的推理能力,适用于多个领域:

规划:其“链式思维”逻辑有助于处理需要逐步推理的任务,特别适合智能体(agentic)应用。

编程:可用于代码生成、调试和优化,提升开发效率。

数学问题求解:具备处理复杂数学问题的能力,在科研、工程和科学计算等领域表现突出。

开发者可在 watsonx.ai 上利用 DeepSeek-R1 进行 AI 解决方案开发,并使用以下解决方案能力:

直观测试和评估模型输出

通过连接向量数据库和嵌入模型构建 RAG(检索增强生成)管道

支持 LangChain、CrewAI 等主流 AI 框架

IBM watsonx.ai 提供灵活的开源模型定制选项,支持 DeepSeek-R1 在不同环境下的部署,并简化智能体开发、微调、RAG、提示工程等工作流。此外,watsonx.ai 内置安全机制,保障用户应用的安全性。

据IT之家此前报道,IBM 首席执行官本月初曾发表长文称,他们(DeepSeek)仅用约 2000 枚英伟达芯片,以大约 600 万美元的成本训练了最新的模型,远低于行业预期。这再次证明,小型、高效的模型也能带来实际成果,无需依赖庞大而昂贵的专有系统。

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Most Discussed

  1. 1
     
     
     
     
  2. 2
     
     
     
     
  3. 3
     
     
     
     
  4. 4
     
     
     
     
  5. 5
     
     
     
     
  6. 6
     
     
     
     
  7. 7
     
     
     
     
  8. 8
     
     
     
     
  9. 9
     
     
     
     
  10. 10