Update README.md
Browse files
README.md
CHANGED
|
@@ -30,6 +30,31 @@ Finetune on [mesolitica/Malaysian-SFT](https://huggingface.co/datasets/mesolitic
|
|
| 30 |
|
| 31 |
Source code at https://github.com/mesolitica/malaya/tree/master/session/gemma3
|
| 32 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
## Acknowledgement
|
| 34 |
|
| 35 |
Special thanks to https://www.sns.com.my for 8x H100 node!
|
|
|
|
| 30 |
|
| 31 |
Source code at https://github.com/mesolitica/malaya/tree/master/session/gemma3
|
| 32 |
|
| 33 |
+
## Benchmark
|
| 34 |
+
|
| 35 |
+
### MalayMMLU
|
| 36 |
+
|
| 37 |
+
Based on 0-shot first token accuracy,
|
| 38 |
+
|
| 39 |
+
```
|
| 40 |
+
Model Accuracy shot by_letter category
|
| 41 |
+
0 Malaysian-gemma-3-1b-it 48.096603 0shot True STEM
|
| 42 |
+
1 Malaysian-gemma-3-1b-it 47.423664 0shot True Language
|
| 43 |
+
2 Malaysian-gemma-3-1b-it 47.210176 0shot True Social science
|
| 44 |
+
3 Malaysian-gemma-3-1b-it 47.709283 0shot True Others
|
| 45 |
+
4 Malaysian-gemma-3-1b-it 51.786121 0shot True Humanities
|
| 46 |
+
{'Social science': 6918, 'Language': 6288, 'Humanities': 4395, 'Others': 4169, 'STEM': 2443}
|
| 47 |
+
Model : Malaysian-gemma-3-1b-it
|
| 48 |
+
Metric : first
|
| 49 |
+
Shot : 0shot
|
| 50 |
+
average accuracy 48.27158964192789
|
| 51 |
+
accuracy for STEM 48.09660253786328
|
| 52 |
+
accuracy for Language 47.4236641221374
|
| 53 |
+
accuracy for Social science 47.21017635154669
|
| 54 |
+
accuracy for Others 47.70928280163108
|
| 55 |
+
accuracy for Humanities 51.786120591581344
|
| 56 |
+
```
|
| 57 |
+
|
| 58 |
## Acknowledgement
|
| 59 |
|
| 60 |
Special thanks to https://www.sns.com.my for 8x H100 node!
|