The file content has been truncated to 1024 tokens.
Generated using ntptoolkit's ntp-training-data and instruction_tuning.py.
It used the following config for ntp-training-data:
{
"repo": "
https://github.com/leanprover-community/mathlib4",
"commit":… See the full description on the dataset page:
https://huggingface.co/datasets/l3lab/ntp-mathlib-instruct-context-fullproof.