[GH-ISSUE #124] Verify evals on Papers with Code #112

Open
opened 2026-10-26 21:08:00 +00:00 by bsakaguchi · 0 comments
Owner

Originally created by @NielsRogge on GitHub (Sep 14, 2026).
Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-V2/issues/124

Hi,

Niels here from the open-source team at Hugging Face. Congratulations on your work!

I've made the paper and 13 paper-native evaluations available on Papers with Code.

The paper has results on Coding Agents, Instruction Following, World Knowledge, and 3 additional task pages.

The DeepSeek-V2 Chat (RL) (5-shot) result currently ranks third on TriviaQA.

The DeepSeek-V2 Chat (RL) result currently ranks fourth on MTbench.

Would it be possible to verify these results and let me know if any score, model name, benchmark protocol, or openness metadata should be corrected?

You can also edit the task, methods, project page, and GitHub URL directly from the paper page using your Hugging Face account.

If you'd like to showcase the results in your repository README, you can copy these live leaderboard badges (or use the “Copy PwC badge” button in the Results section):

Papers with Code: #3 on TriviaQA

Kind regards,

Niels

Originally created by @NielsRogge on GitHub (Sep 14, 2026). Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-V2/issues/124 Hi, Niels here from the open-source team at Hugging Face. Congratulations on your work! I've made the [paper](https://paperswithcode.co/paper/2405.04434) and [13 paper-native evaluations](https://paperswithcode.co/paper/2405.04434#results) available on Papers with Code. The paper has results on [Coding Agents](https://paperswithcode.co/tasks/coding-agents), [Instruction Following](https://paperswithcode.co/tasks/instruction-following), [World Knowledge](https://paperswithcode.co/tasks/world-knowledge), and 3 additional task pages. The DeepSeek-V2 Chat (RL) (5-shot) result currently ranks third on [TriviaQA](https://paperswithcode.co/benchmark/triviaqa?task=world-knowledge&eval=6057). The DeepSeek-V2 Chat (RL) result currently ranks fourth on [MTbench](https://paperswithcode.co/benchmark/mtbench?task=instruction-following&eval=6069). Would it be possible to verify these results and let me know if any score, model name, benchmark protocol, or openness metadata should be corrected? You can also edit the task, methods, project page, and GitHub URL directly from the paper page using your Hugging Face account. If you'd like to showcase the results in your repository README, you can copy these live leaderboard badges (or use the “Copy PwC badge” button in the Results section): [![Papers with Code: #3 on TriviaQA](https://paperswithcode.co/api/v1/papers/2405.04434/leaderboard-badge.svg?eval=6057&live=1)](https://paperswithcode.co/api/v1/papers/2405.04434/leaderboard-badge-link?eval=6057) Kind regards, Niels
Sign in to join this conversation.
No labels
pull-request
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-V2#112
No description provided.