rungalileo/mmhs150k
收藏Hugging Face2026-04-17 更新2026-04-26 收录
下载链接：
https://hf-mirror.com/datasets/rungalileo/mmhs150k
下载链接
链接失效反馈官方服务：
资源简介：
---
configs:
- config_name: default
data_files:
- split: train
path: samples.parquet
---
# MMHS150K Remaining Samples
This dataset contains the MMHS150K samples that were not used in the 3,000-sample manual annotation set.
Rows: 148,623
Columns:
- `input`: ordered list of content blocks. Each block has `type` and `content`. MMHS150K rows contain tweet text followed by an image file reference.
- `original_label`: original MMHS150K `labels_str` annotator labels.
Media files are stored in `multimodal_files.zip`; image content values in `input` correspond to filenames inside that zip.
提供机构：
rungalileo
