Commit
•
d8b9180
1
Parent(s):
5043722
Update README.md
Browse files
README.md
CHANGED
|
@@ -11,12 +11,6 @@ library_name: transformers
|
|
| 11 |
This repository contains a compressed version of the Meta Llama-3-8B-Instruct model, utilizing the Palu framework for KV-Cache compression. Palu reduces the hidden dimensions of the KV-Cache through low-rank decomposition, significantly reducing the model's memory footprint while maintaining or enhancing performance.
|
| 12 |
|
| 13 |
|
| 14 |
-
## Evaluation Results
|
| 15 |
-
|
| 16 |
-
Here's a Markdown file to include the results of your comparisons:
|
| 17 |
-
|
| 18 |
-
---
|
| 19 |
-
|
| 20 |
# Meta Llama-3-8B-Instruct: Palu Compression Results
|
| 21 |
|
| 22 |
## Perplexity (PPL)
|
|
|
|
| 11 |
This repository contains a compressed version of the Meta Llama-3-8B-Instruct model, utilizing the Palu framework for KV-Cache compression. Palu reduces the hidden dimensions of the KV-Cache through low-rank decomposition, significantly reducing the model's memory footprint while maintaining or enhancing performance.
|
| 12 |
|
| 13 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
# Meta Llama-3-8B-Instruct: Palu Compression Results
|
| 15 |
|
| 16 |
## Perplexity (PPL)
|