File size: 501 Bytes
f72c7cc fc1f665 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 |
---
license: apache-2.0
language:
- en
base_model:
- meta-llama/Llama-2-7b-hf
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
This repo contains a 2:4 sparse version of the LLaMA2-7B model. Trainied with methods from AAAI25 paper [Pruning Large Language Models with Semi-Structural Adaptive Sparse Training](https://arxiv.org/abs/2407.20584).
### Model Description
Same structured as LLaMA2-7B, but weight from linear layer conform to 2:4 sparse pattern.
|