---
title: llava-sae-explanations-5k
canonical_url: "https://www.modelscope.cn/datasets/lmms-lab/llava-sae-explanations-5k"
md_url: "https://www.modelscope.cn/datasets/lmms-lab/llava-sae-explanations-5k.md"
repository: lmms-lab/llava-sae-explanations-5k
last_updated: 2024-11-23
license: "Apache License 2.0"
storage_size: "14 GB"
downloads: 2231
stars: 0
---

# llava-sae-explanations-5k

> llava-sae-explanations-5k - lmms-lab 在 ModelScope 开源的数据集。This is the explanation generated for the first 5k features for sae 131k trained on llava-next-llama3-8B.

lmms-lab/llava-sae-explanations-5k 是 ModelScope 魔搭社区上的数据集，存储大小 14 GB，采用 Apache License 2.0 许可。

- **Repository**: lmms-lab/llava-sae-explanations-5k
- **License**: Apache License 2.0
- **Storage size**: 14 GB
- **Downloads**: 2231
- **Stars**: 0
- **Last updated**: 2024-11-23

Source: https://www.modelscope.cn/datasets/lmms-lab/llava-sae-explanations-5k

---

This is the explanation generated for the first 5k features for sae 131k trained on llava-next-llama3-8B. 

The revised one is the explanation using revised prompt and cached on the lmms-lab/sae-sample-cache-dataset. The legacy one is the old explanations with an old prompt and use the first 15% of the LLaVA-NeXT-Data. 

In our paper, the reported evaluation result are based on the revised ones and feature probing is done on both of the versions