Skip to content

压缩token,速度反而变慢 #2

@yqj0919

Description

@yqj0919
<style> </style>
排名 模型 秒/条(原始) 秒/条(压缩token)  
1 gemini-2.0-flash 8.36秒/条 6.20秒/条 gemini-2.0-flash:提速25.84%,加速比:1.35 倍
2 gemini-2.0-flash-lite 6.15秒/条 5.67秒/条 gemini-2.0-flash-lite:提速7.80%,加速比:1.08倍
3 qwen-flash 5.26秒/条 8.12秒/条 压缩反而变慢
4 gemma3 12b 6.97秒/条 8.45秒/条 压缩反而变慢
5 gemma3 4B 15.8秒/条 11.29秒/条 gemma3 4B:提速 28.54%(加速比 1.40 倍)
6 gemma2 9b 8.21秒/条 2.68秒/条 gemma2 9B:提速 67.36%(加速比 3.06 倍)

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions