# Meta-Llama-3-8B-Instruct-abliterated-v3

> Classic early Llama 3 abliterated model. Still useful on lower-end hardware.


- **Base model:** meta-llama/Meta-Llama-3-8B-Instruct
- **Family:** Llama
- **Size:** 8B
- **Method:** abliteration
- **Approx. VRAM (Q4):** ~5–6 GB
- **License:** Llama 3
- **Quantizations:** GGUF, Transformers
- **Last updated:** 2024-12-01
- **Hugging Face:** https://huggingface.co/failspy/Meta-Llama-3-8B-Instruct-abliterated-v3

## Notes

Classic early Llama 3 abliterated model. Still useful on lower-end hardware.

Canonical page: https://www.wahloon.net/models/failspy-llama-3-8b-abliterated

You are responsible for how you use this model.