starlight-apk/Qwen
1
0
Fork 0
mirror of https://github.com/QwenLM/Qwen.git synced 2026-05-20 08:25:47 +08:00
Code Issues Packages Projects Releases Wiki Activity

update agent benchmarks and add qwen-72b results

Browse Source
This commit is contained in:
兼欣
2023-12-06 12:57:11 +08:00
parent a0a557aad8
commit 7eb9016908
7 changed files with 314 additions and 771 deletions
Download Patch File Download Diff File Expand all files Collapse all files

214
README_FR.md
View File

@@ -1029,22 +1029,28 @@ Nous avons testé les capacités d'appel d'outil du modèle sur notre benchmark
<table>
<tr>
<th colspan="4" align="center">Chinese Tool-Use Benchmark</th>
<th colspan="4" align="center">Chinese Tool-Use Benchmark (Version 20231206)</th>
</tr>
<tr>
<th align="center">Model</th><th align="center">Tool Selection (Acc.↑)</th><th align="center">Tool Input (Rouge-L↑)</th><th align="center">False Positive Error↓</th>
</tr>
<tr>
<td>GPT-4</td><td align="center">95%</td><td align="center">0.90</td><td align="center">15.0%</td>
<td>GPT-4</td><td align="center">98.0%</td><td align="center">0.953</td><td align="center">23.9%</td>
</tr>
<tr>
<td>GPT-3.5</td><td align="center">85%</td><td align="center">0.88</td><td align="center">75.0%</td>
<td>GPT-3.5</td><td align="center">74.5%</td><td align="center">0.807</td><td align="center">80.6%</td>
</tr>
<tr>
<td>Qwen-7B-Chat</td><td align="center">98%</td><td align="center">0.91</td><td align="center">7.3%</td>
<td>Qwen-1_8B-Chat</td><td align="center">85.0%</td><td align="center">0.839</td><td align="center">27.6%</td>
</tr>
<tr>
<td>Qwen-14B-Chat</td><td align="center">98%</td><td align="center">0.93</td><td align="center">2.4%</td>
<td>Qwen-7B-Chat</td><td align="center">95.5%</td><td align="center">0.900</td><td align="center">11.6%</td>
</tr>
<tr>
<td>Qwen-14B-Chat</td><td align="center">96.9%</td><td align="center">0.917</td><td align="center">5.6%</td>
</tr>
<tr>
<td>Qwen-72B-Chat</td><td align="center">98.2%</td><td align="center">0.927</td><td align="center">1.1%</td>
</tr>
</table>
@@ -1054,127 +1060,85 @@ Nous avons observé que Qwen est performant en termes d'exécutabilité du code
<table>
<tr>
<th colspan="4" align="center">Executable Rate of Generated Code (%)</th>
<th colspan="5" align="center">Code Interpreter Benchmark (Version 20231206)</th>
</tr>
<tr>
<th align="center">Model</th><th align="center">Math↑</th><th align="center">Visualization↑</th><th align="center">General↑</th>
<th rowspan="2" align="center">Model</th>
<th colspan="3" align="center">Accuracy of Code Execution Results (%)</th>
<th colspan="1" align="center">Executable Rate of Code (%)</th>
</tr>
<tr>
<td>GPT-4</td><td align="center">91.9</td><td align="center">85.9</td><td align="center">82.8</td>
<th align="center">Math↑</th><th align="center">Visualization-Hard↑</th><th align="center">Visualization-Easy↑</th><th align="center">General↑</th>
</tr>
<tr>
<td>GPT-3.5</td><td align="center">89.2</td><td align="center">65.0</td><td align="center">74.1</td>
<td>GPT-4</td>
<td align="center">82.8</td>
<td align="center">66.7</td>
<td align="center">60.8</td>
<td align="center">82.8</td>
</tr>
<tr>
<td>LLaMA2-7B-Chat</td>
<td align="center">41.9</td>
<td align="center">33.1</td>
<td align="center">24.1 </td>
</tr>
<tr>
<td>LLaMA2-13B-Chat</td>
<td align="center">50.0</td>
<td align="center">40.5</td>
<td align="center">48.3 </td>
</tr>
<tr>
<td>CodeLLaMA-7B-Instruct</td>
<td align="center">85.1</td>
<td align="center">54.0</td>
<td align="center">70.7 </td>
</tr>
<tr>
<td>CodeLLaMA-13B-Instruct</td>
<td align="center">93.2</td>
<td align="center">55.8</td>
<td align="center">74.1 </td>
</tr>
<tr>
<td>InternLM-7B-Chat-v1.1</td>
<td align="center">78.4</td>
<td align="center">44.2</td>
<td align="center">62.1 </td>
</tr>
<tr>
<td>InternLM-20B-Chat</td>
<td align="center">70.3</td>
<td align="center">44.2</td>
<td align="center">65.5 </td>
</tr>
<tr>
<td>Qwen-7B-Chat</td>
<td align="center">82.4</td>
<td align="center">64.4</td>
<td align="center">67.2 </td>
</tr>
<tr>
<td>Qwen-14B-Chat</td>
<td align="center">89.2</td>
<td align="center">84.1</td>
<td align="center">65.5</td>
</tr>
</table>
<table>
<tr>
<th colspan="4" align="center">Accuracy of Code Execution Results (%)</th>
</tr>
<tr>
<th align="center">Model</th><th align="center">Math↑</th><th align="center">Visualization-Hard↑</th><th align="center">Visualization-Easy↑</th>
</tr>
<tr>
<td>GPT-4</td><td align="center">82.8</td><td align="center">66.7</td><td align="center">60.8</td>
</tr>
<tr>
<td>GPT-3.5</td><td align="center">47.3</td><td align="center">33.3</td><td align="center">55.7</td>
</tr>
<tr>
<td>LLaMA2-7B-Chat</td>
<td align="center">3.9</td>
<td align="center">14.3</td>
<td align="center">39.2 </td>
<td>GPT-3.5</td>
<td align="center">47.3</td>
<td align="center">33.3</td>
<td align="center">55.7</td>
<td align="center">74.1</td>
</tr>
<tr>
<td>LLaMA2-13B-Chat</td>
<td align="center">8.3</td>
<td align="center">8.3</td>
<td align="center">40.5 </td>
</tr>
<tr>
<td>CodeLLaMA-7B-Instruct</td>
<td align="center">14.3</td>
<td align="center">26.2</td>
<td align="center">60.8 </td>
<td align="center">1.2</td>
<td align="center">15.2</td>
<td align="center">48.3</td>
</tr>
<tr>
<td>CodeLLaMA-13B-Instruct</td>
<td align="center">28.2</td>
<td align="center">27.4</td>
<td align="center">62.0 </td>
</tr>
<tr>
<td>InternLM-7B-Chat-v1.1</td>
<td align="center">28.5</td>
<td align="center">4.8</td>
<td align="center">40.5 </td>
<td align="center">15.5</td>
<td align="center">21.5</td>
<td align="center">74.1</td>
</tr>
<tr>
<td>InternLM-20B-Chat</td>
<td align="center">34.6</td>
<td align="center">10.7</td>
<td align="center">25.1</td>
<td align="center">65.5</td>
</tr>
<tr>
<td>ChatGLM3-6B</td>
<td align="center">54.2</td>
<td align="center">15.5</td>
<td align="center">21.5</td>
<td align="center">67.1</td>
</tr>
<tr>
<td>Qwen-1.8B-Chat</td>
<td align="center">25.6</td>
<td align="center">21.4</td>
<td align="center">45.6 </td>
<td align="center">22.8</td>
<td align="center">65.5</td>
</tr>
<tr>
<td>Qwen-7B-Chat</td>
<td align="center">41.9</td>
<td align="center">40.5</td>
<td align="center">54.4 </td>
<td align="center">23.8</td>
<td align="center">38.0</td>
<td align="center">67.2</td>
</tr>
<tr>
<td>Qwen-14B-Chat</td>
<td align="center">58.4</td>
<td align="center">53.6</td>
<td align="center">59.5</td>
<td align="center">31.0</td>
<td align="center">45.6</td>
<td align="center">65.5</td>
</tr>
<tr>
<td>Qwen-72B-Chat</td>
<td align="center">72.7</td>
<td align="center">41.7</td>
<td align="center">43.0</td>
<td align="center">82.8</td>
</tr>
</table>
@@ -1184,62 +1148,6 @@ Nous avons observé que Qwen est performant en termes d'exécutabilité du code
<br>
<p>
En outre, nous fournissons également des résultats expérimentaux démontrant que notre modèle est capable d'agir en tant qu'agent Hugging Face. Pour plus d'informations, veuillez vous référer à la [documentation de l'exemple](examples/transformers_agent.md). Les performances du modèle sur l'ensemble des données d'évaluation fournies par Hugging Face sont les suivantes:
<table>
<tr>
<th colspan="4" align="center">HuggingFace Agent Benchmark- Run Mode</th>
</tr>
<tr>
<th align="center">Model</th><th align="center">Tool Selection↑</th><th align="center">Tool Used↑</th><th align="center">Code↑</th>
</tr>
<tr>
<td>GPT-4</td><td align="center">100</td><td align="center">100</td><td align="center">97.4</td>
</tr>
<tr>
<td>GPT-3.5</td><td align="center">95.4</td><td align="center">96.3</td><td align="center">87.0</td>
</tr>
<tr>
<td>StarCoder-Base-15B</td><td align="center">86.1</td><td align="center">87.0</td><td align="center">68.9</td>
</tr>
<tr>
<td>StarCoder-15B</td><td align="center">87.0</td><td align="center">88.0</td><td align="center">68.9</td>
</tr>
<tr>
<td>Qwen-7B-Chat</td><td align="center">87.0</td><td align="center">87.0</td><td align="center">71.5</td>
</tr>
<tr>
<td>Qwen-14B-Chat</td><td align="center">93.5</td><td align="center">94.4</td><td align="center">87.0</td>
</tr>
</table>
<table>
<tr>
<th colspan="4" align="center">HuggingFace Agent Benchmark - Chat Mode</th>
</tr>
<tr>
<th align="center">Model</th><th align="center">Tool Selection↑</th><th align="center">Tool Used↑</th><th align="center">Code↑</th>
</tr>
<tr>
<td>GPT-4</td><td align="center">97.9</td><td align="center">97.9</td><td align="center">98.5</td>
</tr>
<tr>
<td>GPT-3.5</td><td align="center">97.3</td><td align="center">96.8</td><td align="center">89.6</td>
</tr>
<tr>
<td>StarCoder-Base-15B</td><td align="center">97.9</td><td align="center">97.9</td><td align="center">91.1</td>
</tr>
<tr>
<td>StarCoder-15B</td><td align="center">97.9</td><td align="center">97.9</td><td align="center">89.6</td>
</tr>
<tr>
<td>Qwen-7B-Chat</td><td align="center">94.7</td><td align="center">94.7</td><td align="center">85.1</td>
</tr>
<tr>
<td>Qwen-14B-Chat</td><td align="center">97.9</td><td align="center">97.9</td><td align="center">95.5</td>
</tr>
</table>
<br>
## Compréhension du Contexte Long