| # 混合 SFT 数据配置（采样版） | |
| # 自动生成 by sample_sft_data.py | |
| # Image: 738,590 条 | |
| # Video: 800,001 条 | |
| # 总计: 1,538,591 条 | |
| datasets: | |
| # ===== Image ===== | |
| - json_path: /mnt/bn/strategy-mllm-train/common/datasets/image_sft_full_v800k.json | |
| sampling_strategy: all | |
| # ===== Video ===== | |
| - json_path: /mnt/bn/strategy-mllm-train/common/datasets/video_sft_full_v800k.json | |
| sampling_strategy: all | |