Upload README.md
Browse files
README.md
CHANGED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
license: other
|
| 3 |
license_name: katanemo-research
|
| 4 |
license_link: >-
|
| 5 |
-
https://huggingface.co/katanemolabs/Arch-Function-1.5B/blob/main/LICENSE
|
| 6 |
base_model:
|
| 7 |
-
-
|
| 8 |
language:
|
| 9 |
- en
|
| 10 |
pipeline_tag: text-generation
|
|
@@ -54,7 +54,7 @@ Katanemo Arch-Function collection is built on top of the [Qwen 2.5](https://hugg
|
|
| 54 |
|
| 55 |
|
| 56 |
## Performance Benchmarks
|
| 57 |
-
We evaluate Katanemo Arch-Function series on the [Berkeley Function-Calling Leaderboard (BFCL)](https://gorilla.cs.berkeley.edu/leaderboard.html#leaderboard).
|
| 58 |
|
| 59 |
<table>
|
| 60 |
<tr style="text-align: center; vertical-align: middle; font-weight: bold;">
|
|
@@ -84,27 +84,16 @@ We evaluate Katanemo Arch-Function series on the [Berkeley Function-Calling Lead
|
|
| 84 |
<td>63.41%</td>
|
| 85 |
<td>82.93%</td>
|
| 86 |
</tr>
|
| 87 |
-
<tr style="text-align: center; vertical-align: middle;">
|
| 88 |
-
<td>
|
| 89 |
-
<td>
|
| 90 |
-
<td>62
|
| 91 |
-
<td>
|
| 92 |
-
<td>
|
| 93 |
-
<td>
|
| 94 |
-
<td>
|
| 95 |
-
<td>
|
| 96 |
-
<td>73.
|
| 97 |
-
</tr>
|
| 98 |
-
<tr style="text-align: center; vertical-align: middle;">
|
| 99 |
-
<td>5</td>
|
| 100 |
-
<td>ToolACE-8B (FC)</td>
|
| 101 |
-
<td>60.44%</td>
|
| 102 |
-
<td>87.06%</td>
|
| 103 |
-
<td>89.52%</td>
|
| 104 |
-
<td>74.99%</td>
|
| 105 |
-
<td>17.38%</td>
|
| 106 |
-
<td>80.49%</td>
|
| 107 |
-
<td>85.71%</td>
|
| 108 |
</tr>
|
| 109 |
<tr style="text-align: center; vertical-align: middle;">
|
| 110 |
<td>6</td>
|
|
@@ -117,28 +106,6 @@ We evaluate Katanemo Arch-Function series on the [Berkeley Function-Calling Lead
|
|
| 117 |
<td>73.17%</td>
|
| 118 |
<td>74.60%</td>
|
| 119 |
</tr>
|
| 120 |
-
<tr style="text-align: center; vertical-align: middle; font-weight: bold;">
|
| 121 |
-
<td> </td>
|
| 122 |
-
<td>Arch-Function-7B</td>
|
| 123 |
-
<td>58.44%</td>
|
| 124 |
-
<td>85.58%</td>
|
| 125 |
-
<td>88.14%</td>
|
| 126 |
-
<td>69.08%</td>
|
| 127 |
-
<td>20.50%</td>
|
| 128 |
-
<td>92.68%</td>
|
| 129 |
-
<td>74.05%</td>
|
| 130 |
-
</tr>
|
| 131 |
-
<tr style="text-align: center; vertical-align: middle; ">
|
| 132 |
-
<td>8</td>
|
| 133 |
-
<td>xLAM-8x22b-r (FC)</td>
|
| 134 |
-
<td>57.99%</td>
|
| 135 |
-
<td>88.15%</td>
|
| 136 |
-
<td>90.11%</td>
|
| 137 |
-
<td>71.97%</td>
|
| 138 |
-
<td>14.50%</td>
|
| 139 |
-
<td>85.37%</td>
|
| 140 |
-
<td>67.29%</td>
|
| 141 |
-
</tr>
|
| 142 |
<tr style="text-align: center; vertical-align: middle; ">
|
| 143 |
<td>9</td>
|
| 144 |
<td>Gemini-1.5-Flash-002 (Prompt)</td>
|
|
@@ -150,16 +117,16 @@ We evaluate Katanemo Arch-Function series on the [Berkeley Function-Calling Lead
|
|
| 150 |
<td>85.37%</td>
|
| 151 |
<td>78.54%</td>
|
| 152 |
</tr>
|
| 153 |
-
<tr style="text-align: center; vertical-align: middle; ">
|
| 154 |
-
<td>
|
| 155 |
-
<td>
|
| 156 |
<td>57.69%</td>
|
| 157 |
-
<td>
|
| 158 |
-
<td>
|
| 159 |
-
<td>
|
| 160 |
-
<td>
|
| 161 |
-
<td>
|
| 162 |
-
<td>
|
| 163 |
</tr>
|
| 164 |
<tr style="text-align: center; vertical-align: middle; ">
|
| 165 |
<td>12</td>
|
|
@@ -183,50 +150,16 @@ We evaluate Katanemo Arch-Function series on the [Berkeley Function-Calling Lead
|
|
| 183 |
<td>75.61%</td>
|
| 184 |
<td>49.44%</td>
|
| 185 |
</tr>
|
| 186 |
-
<tr style="text-align: center; vertical-align: middle; font-weight: bold;">
|
| 187 |
-
<td> </td>
|
| 188 |
-
<td>Arch-Function-3B</td>
|
| 189 |
-
<td>56.57%</td>
|
| 190 |
-
<td>83.62%</td>
|
| 191 |
-
<td>85.36%</td>
|
| 192 |
-
<td>66.90%</td>
|
| 193 |
-
<td>19.50%</td>
|
| 194 |
-
<td>97.56%</td>
|
| 195 |
-
<td>70.99%</td>
|
| 196 |
-
</tr>
|
| 197 |
-
</tr>
|
| 198 |
<tr style="text-align: center; vertical-align: middle; font-weight: bold;">
|
| 199 |
<td> </td>
|
| 200 |
<td>Arch-Function-1.5B</td>
|
| 201 |
-
<td>
|
| 202 |
-
<td>
|
| 203 |
-
<td>
|
| 204 |
-
<td>
|
| 205 |
-
<td>
|
| 206 |
-
<td>
|
| 207 |
-
<td>
|
| 208 |
-
</tr>
|
| 209 |
-
<tr style="text-align: center; vertical-align: middle; ">
|
| 210 |
-
<td>19</td>
|
| 211 |
-
<td>xLAM-7b-r (FC)</td>
|
| 212 |
-
<td>54.41%</td>
|
| 213 |
-
<td>81.40%</td>
|
| 214 |
-
<td>83.46%</td>
|
| 215 |
-
<td>67.88%</td>
|
| 216 |
-
<td>14.50%</td>
|
| 217 |
-
<td>97.56%</td>
|
| 218 |
-
<td>64.05%</td>
|
| 219 |
-
</tr>
|
| 220 |
-
<tr style="text-align: center; vertical-align: middle; ">
|
| 221 |
-
<td>20</td>
|
| 222 |
-
<td>Qwen2.5-7B-Instruct (Prompt)</td>
|
| 223 |
-
<td>54.27%</td>
|
| 224 |
-
<td>85.79%</td>
|
| 225 |
-
<td>88.13%</td>
|
| 226 |
-
<td>65.97%</td>
|
| 227 |
-
<td>11.25%</td>
|
| 228 |
-
<td>92.68%</td>
|
| 229 |
-
<td>64.95%</td>
|
| 230 |
</tr>
|
| 231 |
<tr style="text-align: center; vertical-align: middle; ">
|
| 232 |
<td>21</td>
|
|
@@ -409,4 +342,4 @@ The current temperature in Seattle is 62 degrees in Fahrenheit.
|
|
| 409 |
|
| 410 |
|
| 411 |
# License
|
| 412 |
-
Katanemo Arch-Function collection is distributed under the [Katanemo license](https://huggingface.co/katanemolabs/Arch-Function-1.5B/blob/main/LICENSE).
|
|
|
|
| 2 |
license: other
|
| 3 |
license_name: katanemo-research
|
| 4 |
license_link: >-
|
| 5 |
+
https://huggingface.co/katanemolabs/Arch-Function-1.5B.gguf/blob/main/LICENSE
|
| 6 |
base_model:
|
| 7 |
+
- katanemo/Arch-Function-1.5B
|
| 8 |
language:
|
| 9 |
- en
|
| 10 |
pipeline_tag: text-generation
|
|
|
|
| 54 |
|
| 55 |
|
| 56 |
## Performance Benchmarks
|
| 57 |
+
We evaluate Katanemo Arch-Function series on the [Berkeley Function-Calling Leaderboard (BFCL)](https://gorilla.cs.berkeley.edu/leaderboard.html#leaderboard). We compare with commonly-used models and the results (as of Oct 21st, 2024) are shwon below. For each model family, we select the one with the highest rank.
|
| 58 |
|
| 59 |
<table>
|
| 60 |
<tr style="text-align: center; vertical-align: middle; font-weight: bold;">
|
|
|
|
| 84 |
<td>63.41%</td>
|
| 85 |
<td>82.93%</td>
|
| 86 |
</tr>
|
| 87 |
+
<tr style="text-align: center; vertical-align: middle; font-weight: bold;">
|
| 88 |
+
<td> </td>
|
| 89 |
+
<td>Arch-Function-7B</td>
|
| 90 |
+
<td>59.62%</td>
|
| 91 |
+
<td>86.83%</td>
|
| 92 |
+
<td>88.07%</td>
|
| 93 |
+
<td>71.57%</td>
|
| 94 |
+
<td>21.00%</td>
|
| 95 |
+
<td>95.12%</td>
|
| 96 |
+
<td>73.63%</td>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 97 |
</tr>
|
| 98 |
<tr style="text-align: center; vertical-align: middle;">
|
| 99 |
<td>6</td>
|
|
|
|
| 106 |
<td>73.17%</td>
|
| 107 |
<td>74.60%</td>
|
| 108 |
</tr>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 109 |
<tr style="text-align: center; vertical-align: middle; ">
|
| 110 |
<td>9</td>
|
| 111 |
<td>Gemini-1.5-Flash-002 (Prompt)</td>
|
|
|
|
| 117 |
<td>85.37%</td>
|
| 118 |
<td>78.54%</td>
|
| 119 |
</tr>
|
| 120 |
+
<tr style="text-align: center; vertical-align: middle; font-weight: bold;">
|
| 121 |
+
<td> </td>
|
| 122 |
+
<td>Arch-Function-3B</td>
|
| 123 |
<td>57.69%</td>
|
| 124 |
+
<td>85.19%</td>
|
| 125 |
+
<td>86.18%</td>
|
| 126 |
+
<td>71.21%</td>
|
| 127 |
+
<td>17.50%</td>
|
| 128 |
+
<td>90.24%</td>
|
| 129 |
+
<td>72.88%</td>
|
| 130 |
</tr>
|
| 131 |
<tr style="text-align: center; vertical-align: middle; ">
|
| 132 |
<td>12</td>
|
|
|
|
| 150 |
<td>75.61%</td>
|
| 151 |
<td>49.44%</td>
|
| 152 |
</tr>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 153 |
<tr style="text-align: center; vertical-align: middle; font-weight: bold;">
|
| 154 |
<td> </td>
|
| 155 |
<td>Arch-Function-1.5B</td>
|
| 156 |
+
<td>56.20%</td>
|
| 157 |
+
<td>84.40%</td>
|
| 158 |
+
<td>83.96%</td>
|
| 159 |
+
<td>69.36%</td>
|
| 160 |
+
<td>15.88%</td>
|
| 161 |
+
<td>87.80%</td>
|
| 162 |
+
<td>74.39%</td>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 163 |
</tr>
|
| 164 |
<tr style="text-align: center; vertical-align: middle; ">
|
| 165 |
<td>21</td>
|
|
|
|
| 342 |
|
| 343 |
|
| 344 |
# License
|
| 345 |
+
Katanemo Arch-Function collection is distributed under the [Katanemo license](https://huggingface.co/katanemolabs/Arch-Function-1.5B.gguf/blob/main/LICENSE).
|