DeepSeek Found Using SmolVLM to Curate Pretraining Image-Text Data

Hugging Face researcher Elie Bakouch discovered that DeepSeek used SmolVLM to strictly score and curate image-text pretraining data, noting with pride that his team's model was adopted in industry. He later clarified the finding came from an earlier DeepSeek technical report.

2026-09-10 ~ 2026-09-10 · 3 related posts

• SmolVLM used for strict image-text quality scoring in new model tech report — eliebakouch · 2026-09-10
• DeepSeek used SmolVLM to quality-filter interleaved pretraining data, researcher spots — eliebakouch · 2026-09-10
• Correction: the SmolVLM data-filtering trick is from DeepSeek's earlier tech report — eliebakouch · 2026-09-10