Version of the nvidia/Nemotron-SFT-Instruction-Following-Chat-v2 dataset used in the AMALIA's Supervised Fine-Tuning stage, both in the base and ramp down stages. The ramp down stage comprised a randomly select subset of the base subset, where part was translated to European Portuguese using google/gemma-4-31B-it.
This dataset went through a processing pipeline to:
Remove entries that reference other LLMs or research labs;… See the full description on the dataset page:
https://huggingface.co/datasets/amalia-llm/amalia-Nemotron-SFT-Instruction-Following-Chat-v2.
