Update README.md
Browse files
README.md
CHANGED
|
@@ -80,30 +80,42 @@ The github link is: [LumenScopeAI/BrainTransformers-SNN-LLM](https://github.com/
|
|
| 80 |
|
| 81 |
Below are the performance metrics of our 3B model on various benchmarks:
|
| 82 |
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
|
| 86 |
-
|
|
| 87 |
-
|
|
| 88 |
-
|
|
| 89 |
-
|
|
| 90 |
-
|
|
| 91 |
-
|
|
| 92 |
-
|
|
| 93 |
-
|
|
| 94 |
-
|
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
|
| 98 |
-
|
|
| 99 |
-
|
|
| 100 |
-
|
|
| 101 |
-
|
|
| 102 |
-
|
|
| 103 |
-
|
|
| 104 |
-
|
|
| 105 |
-
|
| 106 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 107 |
|
| 108 |
## Usage
|
| 109 |
|
|
|
|
| 80 |
|
| 81 |
Below are the performance metrics of our 3B model on various benchmarks:
|
| 82 |
|
| 83 |
+
### General Tasks
|
| 84 |
+
|
| 85 |
+
| Dataset | Performance |
|
| 86 |
+
|---------|-------------|
|
| 87 |
+
| MMLU | 63.2 |
|
| 88 |
+
| MMLU-pro | 33.3 |
|
| 89 |
+
| MMLU-redux | 61.3 |
|
| 90 |
+
| BBH | 54.1 |
|
| 91 |
+
| ARC-C | 54.3 |
|
| 92 |
+
| Trurhfulqa | 47.1 |
|
| 93 |
+
| Winogrande | 68.8 |
|
| 94 |
+
| Hellaswag | 72.8 |
|
| 95 |
+
|
| 96 |
+
### Math and Science Tasks
|
| 97 |
+
|
| 98 |
+
| Dataset | Performance |
|
| 99 |
+
|---------|-------------|
|
| 100 |
+
| GPQA | 25.3 |
|
| 101 |
+
| Theoremqa | 26.4 |
|
| 102 |
+
| MATH | 41.0 |
|
| 103 |
+
| MMLU-stem | 60.2 |
|
| 104 |
+
| GSM8K | 76.3 |
|
| 105 |
+
|
| 106 |
+
### Coding and Multilingual Tasks
|
| 107 |
+
|
| 108 |
+
| Dataset | Performance |
|
| 109 |
+
|---------|-------------|
|
| 110 |
+
| HumanEval | 40.5 |
|
| 111 |
+
| HumanEval+ | 34.6 |
|
| 112 |
+
| MBPP | 55.0 |
|
| 113 |
+
| MBPP+ | 47.5 |
|
| 114 |
+
| MultiPL-E | 39.6 |
|
| 115 |
+
| Multi-Exam | 52.6 |
|
| 116 |
+
| Multi-Understanding | 73.9 |
|
| 117 |
+
| Multi-Mathematics | 47.1 |
|
| 118 |
+
| Multi-Translation | 28.2 |
|
| 119 |
|
| 120 |
## Usage
|
| 121 |
|