ultimate-code is a derived dataset built by combining and processing data from
nvidia/OpenCodeInstruct and
nvidia/OpenCodeGeneticInstruct.
It can be used to fine-tune LLMs for coding tasks.
Tokenized variant available: a pre-tokenized version of this dataset is available at
CodeForCodersYT/ultimate-code-tokenized.
Use that version if you want ready-to-train tokenized sequences instead of raw text.
Source Datasets &… See the full description on the dataset page: https://huggingface.co/datasets/CodeForCodersYT/ultimate-code-tokenized.
