[GH-ISSUE #520] 请问一下distill model里提供的几个指标大概测了多久 #410

Closed
opened 2026-10-26 21:20:50 +00:00 by bsakaguchi · 3 comments
Owner

Originally created by @BlackTea-c on GitHub (Mar 1, 2025).
Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-R1/issues/520

比如qwen1.5B的
自己下来测其中一个math500感觉都要好几天。
GPTQA 1.5B也要8h

Originally created by @BlackTea-c on GitHub (Mar 1, 2025). Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-R1/issues/520 比如qwen1.5B的 自己下来测其中一个math500感觉都要好几天。 GPTQA 1.5B也要8h
Author
Owner

@BugMyGod commented on GitHub (Mar 1, 2025):

这是来自QQ邮箱的假期自动回复邮件。
 
您好,我最近正在休假中,无法亲自回复您的邮件。我将在假期结束后,尽快给您回复。

<!-- gh-comment-id:2692346924 --> @BugMyGod commented on GitHub (Mar 1, 2025): 这是来自QQ邮箱的假期自动回复邮件。   您好,我最近正在休假中,无法亲自回复您的邮件。我将在假期结束后,尽快给您回复。
Author
Owner

@ice-tong commented on GitHub (Mar 19, 2025):

请问有测过 deepseek-r1-distill-qwen-1.5b 吗,我们这边测试发现 ppl@wikitext2 高达 50+,有点太大了吧。

<!-- gh-comment-id:2735641059 --> @ice-tong commented on GitHub (Mar 19, 2025): 请问有测过 deepseek-r1-distill-qwen-1.5b 吗,我们这边测试发现 ppl@wikitext2 高达 50+,有点太大了吧。
Author
Owner

@BlackTea-c commented on GitHub (Mar 28, 2025):

vllm

<!-- gh-comment-id:2760071428 --> @BlackTea-c commented on GitHub (Mar 28, 2025): vllm
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
deepseek-ai/DeepSeek-R1#410
No description provided.