---
library_name: saelens
license: apache-2.0
datasets:
- Juliushanhanhan/openwebtext-1b-llama3-tokenized-cxt-1024
---

# Llama-3-8B SAEs (layer 25, Post-MLP Residual Stream)

## Introduction

We train a Gated SAE on the post-MLP residual stream of the 25th layer of [Llama-3-8b-instruct](https://huggingface.co./meta-llama/Meta-Llama-3-8B-Instruct) model. The width of SAE hidden dimensions is 65536 (x16). 

The SAE is trained with 500M tokens from the [OpenWebText corpus](https://huggingface.co./datasets/Juliushanhanhan/openwebtext-1b-llama3-tokenized-cxt-1024).

Feature visualizations are hosted at https://www.neuronpedia.org/llama3-8b-it. The wandb run is recorded [here](https://wandb.ai/jiatongg/sae_semantic_entropy/runs/ruuu0izg?nw=nwuserjiatongg).

## Load the Model


This repository contains the following SAEs:
- blocks.25.hook_resid_post

Load these SAEs using SAELens as below:
```python
from sae_lens import SAE

sae, cfg_dict, sparsity = SAE.from_pretrained("Juliushanhanhan/llama-3-8b-it-res", "<sae_id>")
```

## Citation

```
@misc{saelens2024llama38b,
  author = {SAELens, Jiatong Han},
  title = {Llama-3-8B SAEs (layer 25, Post-MLP Residual Stream)},
  year = {2024},
  publisher = {HuggingFace},
  url = {https://huggingface.co./Juliushanhanhan/llama-3-8b-it-res},
  note = {Model trained on the post-MLP residual stream of the 25th layer of Llama-3-8B. Feature visualizations are available at \url{https://www.neuronpedia.org/llama3-8b-it}. The wandb run is recorded at \url{https://wandb.ai/jiatongg/sae_semantic_entropy/runs/ruuu0izg?nw=nwuserjiatongg}.},
}

@misc{juliushanhanhan2024openwebtext,
  author = {Juliushanhanhan},
  title = {OpenWebText-1B Llama3 Tokenized CXT 1024},
  year = {2024},
  publisher = {HuggingFace},
  url = {https://huggingface.co./datasets/Juliushanhanhan/openwebtext-1b-llama3-tokenized-cxt-1024},
  note = {Dataset used for training the Llama-3-8B SAEs.},
}

```