A dataset of 2,000 humorous one-liners, jokes, and witty responses formatted for instruction-tuned language models (e.g., LLaMA 2, GPT-style).The dataset is designed to help train and fine-tune models that can generate short, punchy, and context-aware humor.
Name: humorchains-llama2-1k
Modality: Text
Size: 2,000 samples (~324 KB)
Format: Instruction-style (
[INST] ... [/INST] ... )
Use Case: Fine-tuning… See the full description on the dataset page:
https://huggingface.co/datasets/anuzb/humorchains-llama2-1k.
