[GH-ISSUE #56] dataset format of pretraining stage #44

Open
opened 2026-10-26 21:09:05 +00:00 by bsakaguchi · 0 comments
Owner

Originally created by @annopackage on GitHub (Jul 17, 2024).
Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-VL/issues/56

How did you unify the format of pretraining dataset? During supervised fine tuning stage, the training data are curated as question and answer pairs. For caption or detection dataset, I want to know if they follow the same format as sft data, and how to collect questions for these data as they originally only contains ground truth like caption or boxes?

Originally created by @annopackage on GitHub (Jul 17, 2024). Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-VL/issues/56 How did you unify the format of pretraining dataset? During supervised fine tuning stage, the training data are curated as question and answer pairs. For caption or detection dataset, I want to know if they follow the same format as sft data, and how to collect questions for these data as they originally only contains ground truth like caption or boxes?
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-VL#44
No description provided.