DeepSeek-R1 · 官方评测披露

DeepSeek-R1 当前收录 27 项测试的官方记录。已完成列明材料的核查。

模型发布 · DeepSeek R1

官方材料与核查范围

All listed official model-card, official News/API, metadata and directly linked technical-report materials for this release were read through the requested scope. Tables without an explicit unit retain an empty unit; figures without exact labels were not estimated. 核对时间 。

当前资料不足以还原首发。 No reliable historical snapshot proving the complete launch-day material set was obtained. Official publication dates and current materials do not substitute for a launch snapshot.

  • DeepSeek-R1 official Hugging Face model card · 主文 · HF revision 56d4cbbb4d29f4355bab4b9a39ccb717a14ad5ad; current card snapshot
    公布 未确认 · 采集 · 已核查 Official model-card Markdown read in full; benchmark rows, notes, configuration statements and linked reports checked.
  • DeepSeek-R1 official Hugging Face model metadata · 主文 · author=deepseek-ai; repo createdAt=2025-01-20T03:46:07.000Z (repository upload time only); lastModified=2025-03-27T04:01:59.000Z; sha=56d4cbbb4d29f4355bab4b9a39ccb717a14ad5ad
    公布 未确认 · 采集 · 已核查 Official model API metadata checked for author, repository identity, revision and timestamps; createdAt was never used as release date.
  • DeepSeek-R1 发布,性能对标 OpenAI o1 正式版 · 主文 · Official DeepSeek News index entry dated 2025-01-20; index downloaded and retained locally
    公布 · 采集 · 已核查 The official dated entry or current service table was read for this release scope; aliases were not treated as weight checkpoints.
  • DeepSeek-R1 paper · 附件 · Official DeepSeek-R1 paper: 86/86 PDF pages read; Table 3, Table 8, Table 13, Table 14 and Table 15 checked. Safety, jailbreak and attack sections pp.44-54 were retained as local audit only and excluded from regular mappings.
    公布 未确认 · 采集 · 已核查 Official DeepSeek-R1 paper: 86/86 PDF pages read; Table 3, Table 8, Table 13, Table 14 and Table 15 checked. Safety, jailbreak and attack sections pp.44-54 were retained as local audit only and excluded from regular mappings.

配置

  • DeepSeek-R1 reported reasoning configuration · Maximum generation length 32,768 tokens. For sampled benchmarks: temperature=0.6, top_p=0.95, 64 responses per query to estimate Pass@1. 输出上限 32768 tokens
    未知项 Per-benchmark sampling use is not itemized beyond the card preface.

直接前代与后续发布

当前收录未确认直接前代。 官方关系依据

测试记录 · 27