high-quality Chinese training datasets
Collection
a suite of high-quality Chinese datasets, used for pretraining, fine-tuning or reinforcement learning.
•
9 items
•
Updated
•
3
- We apply DPO to
opencsg/csg-wukong-2b-smoltalk-chinese, usingultrafeedback-chinese-binarizedas the DPO dataset. - Train 2 epochs, learning rate=5e-6, batchsize=128
- Downloads last month
- 0
Model tree for opencsg/csg-wukong-2b-ultrafeedback-chinese
Base model
opencsg/csg-wukong-2b-chinese-fineweb-edu
Finetuned
opencsg/csg-wukong-2b-smoltalk-chinese