RWKV-Runner?
#2
by
mzeid
- opened
Thanks for sharing the model. Have you used RWKV-Runner to finetune 7B-World model? Do you mind sharing the size of your dataset and how many epochs have you run?
How long did it take you? What is your GPU memory, etc.?
Are you happy with the overall quality?
What is the structure of your training data? 2 rows would be sufficient.
Thanks
Use this repo if you have programing skill: https://github.com/JL-er/RWKV-PEFT
And use RWKV v6 models or mobius rwkv r6 12B which suport function call
if you only have 2 rows, you can repeat the dataset 100 times and shuffle it.