Update README.md
Browse files
README.md
CHANGED
|
@@ -4,12 +4,15 @@ language:
|
|
| 4 |
- ja
|
| 5 |
---
|
| 6 |
|
|
|
|
|
|
|
|
|
|
| 7 |
```
|
| 8 |
import torch
|
| 9 |
from transformers import AutoModelForCausalLM, AutoTokenizer, AwqConfig
|
| 10 |
|
| 11 |
model_id = "dahara1/llama3.1-8b-Instruct-awq"
|
| 12 |
-
|
| 13 |
|
| 14 |
quantization_config = AwqConfig(
|
| 15 |
bits=4,
|
|
|
|
| 4 |
- ja
|
| 5 |
---
|
| 6 |
|
| 7 |
+
量子化時に日本語と中国語を多めに追加しているため、[hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4](hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4)より日本語データを使って計測したPerplexityが良い事がわかっています
|
| 8 |
+
|
| 9 |
+
|
| 10 |
```
|
| 11 |
import torch
|
| 12 |
from transformers import AutoModelForCausalLM, AutoTokenizer, AwqConfig
|
| 13 |
|
| 14 |
model_id = "dahara1/llama3.1-8b-Instruct-awq"
|
| 15 |
+
|
| 16 |
|
| 17 |
quantization_config = AwqConfig(
|
| 18 |
bits=4,
|