Update README.md
Browse files
README.md
CHANGED
|
@@ -13,6 +13,20 @@ tags:
|
|
| 13 |
|
| 14 |
arco lite is a passthrough model based on danube outputs to keep generality, even though its performance decreased, it's stil better than qwen2 at most benchmarks.
|
| 15 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
### Configuration
|
| 17 |
|
| 18 |
The following YAML configuration was used to produce this model:
|
|
|
|
| 13 |
|
| 14 |
arco lite is a passthrough model based on danube outputs to keep generality, even though its performance decreased, it's stil better than qwen2 at most benchmarks.
|
| 15 |
|
| 16 |
+
#### benchmarks
|
| 17 |
+
|
| 18 |
+
zero-shot evaluations performed on current sota ~0.5b models.
|
| 19 |
+
|
| 20 |
+
| Parameters | Model | MMLU | ARC-C | HellaSwag | PIQA | Winogrande | Average |
|
| 21 |
+
| -----------|--------------------------------|-------|-------|-----------|--------|------------|---------|
|
| 22 |
+
| 0.5b | qwen2 |**44.13**| 28.92| 49.05 | 69.31 | 56.99 | 49.68 |
|
| 23 |
+
| 1.1b | tinyllama | 25.77| 30.29| 59.35 | 73.29 | 59.59 | 49.66 |
|
| 24 |
+
| 0.5b | danube3-base | 24.81| 36.18| 60.46| 73.78 | 61.01 | 51.25 |
|
| 25 |
+
| 0.5b | danube3-chat | 25.54 | 36.26 | 60.72 | 74.32 | 61.40 | 51.64 |
|
| 26 |
+
| 0.5b | palmer-004-turbo | 27.36|35.58|61.79|73.67 | 61.17 |51.91|
|
| 27 |
+
| 1.1b | palmer-004 |26.61 | 34.90|61.73 |**74.81**|**64.17**|52.44|
|
| 28 |
+
| 0.5b | arco |26.17|**37.29**|**62.88**|74.37|62.27|**52.60**|
|
| 29 |
+
|
| 30 |
### Configuration
|
| 31 |
|
| 32 |
The following YAML configuration was used to produce this model:
|