Update README: add training and validation info
Browse files
README.md
CHANGED
|
@@ -9,4 +9,103 @@ base_model:
|
|
| 9 |
library_name: transformers
|
| 10 |
tags:
|
| 11 |
- text-generation-inference
|
| 12 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
library_name: transformers
|
| 10 |
tags:
|
| 11 |
- text-generation-inference
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
# QwenMath
|
| 15 |
+
|
| 16 |
+
A generation LLM which can solve math problems.
|
| 17 |
+
|
| 18 |
+
## Training Statistics
|
| 19 |
+
```yaml
|
| 20 |
+
training-method: lora
|
| 21 |
+
training-time: "5:42"
|
| 22 |
+
data-size: 500
|
| 23 |
+
epoch: 3
|
| 24 |
+
total_flos: "1372250GF"
|
| 25 |
+
train_loss: 0.6441
|
| 26 |
+
train_samples_per_second: 4.385
|
| 27 |
+
train_steps_per_second: 0.544
|
| 28 |
+
```
|
| 29 |
+
|
| 30 |
+
## Validation Set Performance
|
| 31 |
+
Dataset used: test split of [fdyrd/MATH](https://huggingface.co/datasets/fdyrd/MATH).
|
| 32 |
+
|
| 33 |
+
<table>
|
| 34 |
+
<tr>
|
| 35 |
+
<th> Level </th>
|
| 36 |
+
<th> Algebra </th>
|
| 37 |
+
<th> Intermediate Algebra </th>
|
| 38 |
+
<th> Prealgebra </th>
|
| 39 |
+
<th> Precalculus </th>
|
| 40 |
+
<th> Number Theory </th>
|
| 41 |
+
<th> Geometry </th>
|
| 42 |
+
<th> Counting & Probability </th>
|
| 43 |
+
<th> Average </th>
|
| 44 |
+
</tr>
|
| 45 |
+
<tr>
|
| 46 |
+
<td> Level 1 </td>
|
| 47 |
+
<td> 0.541 : 135 </td>
|
| 48 |
+
<td> 0.192 : 52 </td>
|
| 49 |
+
<td> 0.477 : 86 </td>
|
| 50 |
+
<td> 0.228 : 57 </td>
|
| 51 |
+
<td> 0.467 : 30 </td>
|
| 52 |
+
<td> 0.263 : 38 </td>
|
| 53 |
+
<td> 0.359 : 39 </td>
|
| 54 |
+
<td> 0.361 </td>
|
| 55 |
+
</tr>
|
| 56 |
+
<tr>
|
| 57 |
+
<td> Level 2 </td>
|
| 58 |
+
<td> 0.323 : 201 </td>
|
| 59 |
+
<td> 0.109 : 128 </td>
|
| 60 |
+
<td> 0.367 : 177 </td>
|
| 61 |
+
<td> 0.044 : 113 </td>
|
| 62 |
+
<td> 0.38 : 92 </td>
|
| 63 |
+
<td> 0.134 : 82 </td>
|
| 64 |
+
<td> 0.248 : 101 </td>
|
| 65 |
+
<td> 0.229 </td>
|
| 66 |
+
</tr>
|
| 67 |
+
<tr>
|
| 68 |
+
<td> Level 3 </td>
|
| 69 |
+
<td> 0.291 : 261 </td>
|
| 70 |
+
<td> 0.046 : 195 </td>
|
| 71 |
+
<td> 0.308 : 224 </td>
|
| 72 |
+
<td> 0.0 : 127 </td>
|
| 73 |
+
<td> 0.262 : 122 </td>
|
| 74 |
+
<td> 0.088 : 102 </td>
|
| 75 |
+
<td> 0.16 : 100 </td>
|
| 76 |
+
<td> 0.165 </td>
|
| 77 |
+
</tr>
|
| 78 |
+
<tr>
|
| 79 |
+
<td> Level 4 </td>
|
| 80 |
+
<td> 0.18 : 283 </td>
|
| 81 |
+
<td> 0.024 : 248 </td>
|
| 82 |
+
<td> 0.22 : 191 </td>
|
| 83 |
+
<td> 0.009 : 114 </td>
|
| 84 |
+
<td> 0.169 : 142 </td>
|
| 85 |
+
<td> 0.064 : 125 </td>
|
| 86 |
+
<td> 0.09 : 111 </td>
|
| 87 |
+
<td> 0.108 </td>
|
| 88 |
+
</tr>
|
| 89 |
+
<tr>
|
| 90 |
+
<td> Level 5 </td>
|
| 91 |
+
<td> 0.088 : 307 </td>
|
| 92 |
+
<td> 0.004 : 280 </td>
|
| 93 |
+
<td> 0.104 : 193 </td>
|
| 94 |
+
<td> 0.0 : 135 </td>
|
| 95 |
+
<td> 0.136 : 154 </td>
|
| 96 |
+
<td> 0.023 : 132 </td>
|
| 97 |
+
<td> 0.065 : 123 </td>
|
| 98 |
+
<td> 0.06 </td>
|
| 99 |
+
</tr>
|
| 100 |
+
<tr>
|
| 101 |
+
<td> Average </td>
|
| 102 |
+
<td> 0.285 </td>
|
| 103 |
+
<td> 0.075 </td>
|
| 104 |
+
<td> 0.295 </td>
|
| 105 |
+
<td> 0.056 </td>
|
| 106 |
+
<td> 0.283 </td>
|
| 107 |
+
<td> 0.114 </td>
|
| 108 |
+
<td> 0.184 </td>
|
| 109 |
+
<td> 0.166 </td>
|
| 110 |
+
</tr>
|
| 111 |
+
</table>
|