Commit
•
35ef64f
1
Parent(s):
7faedd6
Update README.md
Browse files
README.md
CHANGED
|
@@ -9,9 +9,9 @@ base_model:
|
|
| 9 |
- ## Introduction
|
| 10 |
This model was created using Quark Quantization, followed by OGA Model Builder, and finalized with post-processing for NPU deployment.
|
| 11 |
- ## Quantization Strategy
|
| 12 |
-
- AWQ / Group 128 / Asymmetric / BF16 activations
|
| 13 |
- ## Quick Start
|
| 14 |
-
For quickstart, refer to npu-llm-artifacts_1.3.zip available in [RyzenAI-SW-EA](https://account.amd.com/en/member/ryzenai-sw-ea.html)
|
| 15 |
|
| 16 |
#### Evaluation scores
|
| 17 |
The perplexity measurement is run on the wikitext-2-raw-v1 (raw data) dataset provided by Hugging Face. Perplexity score measured for prompt length 2k is 12.27.
|
|
|
|
| 9 |
- ## Introduction
|
| 10 |
This model was created using Quark Quantization, followed by OGA Model Builder, and finalized with post-processing for NPU deployment.
|
| 11 |
- ## Quantization Strategy
|
| 12 |
+
- AWQ / Group 128 / Asymmetric / BF16 activations / UINT4 Weights
|
| 13 |
- ## Quick Start
|
| 14 |
+
For quickstart, refer to npu-llm-artifacts_1.3.0.zip available in [RyzenAI-SW-EA](https://account.amd.com/en/member/ryzenai-sw-ea.html)
|
| 15 |
|
| 16 |
#### Evaluation scores
|
| 17 |
The perplexity measurement is run on the wikitext-2-raw-v1 (raw data) dataset provided by Hugging Face. Perplexity score measured for prompt length 2k is 12.27.
|