Score less than 1.0

[offline888]

Thank you for your excellent work！

When performing inference on SPAQ, sometimes the inference results show scores less than 1.0 .
Is this phenomenon normal? Are there any suitable prompts to avoid it?

[TianheWu]

Thanks. When evaluate the SPAQ, the image could be resized to 768 resolution (short edge).

…

________________________________ 发件人: offline888 ***@***.***> 发送时间: 2026年1月21日 15:52 收件人: TianheWu/VisualQuality-R1 ***@***.***> 抄送: Subscribed ***@***.***> 主题: [Ext] [TianheWu/VisualQuality-R1] Score less than 1.0 (Issue #11) CAUTION: External email. Do not reply, click on links or open attachments unless you recognize the sender and know the content is safe. [https://avatars.githubusercontent.com/u/149365261?s=20&v=4]offline888 created an issue (TianheWu/VisualQuality-R1#11)<#11> Thank you for your excellent work！ When performing inference on SPAQ, sometimes the inference results show scores less than 1.0 . Is this phenomenon normal? Are there any suitable prompts to avoid it? ― Reply to this email directly, view it on GitHub<#11>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/APC47QQ7CA3MGTELGINA5OT4H4V2VAVCNFSM6AAAAACSL2MOY6VHI2DSMVQWIX3LMV43ASLTON2WKOZTHAZTOMBSGM2TEOI>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[offline888]

Yes, I've already resized it, but still got a score less than 1.0

[TianheWu]

Thanks. Is this issue in other datasets?

…

________________________________ 发件人: offline888 ***@***.***> 发送时间: 2026年1月21日 16:33 收件人: TianheWu/VisualQuality-R1 ***@***.***> 抄送: WU Tianhe ***@***.***>; Comment ***@***.***> 主题: [Ext] Re: [TianheWu/VisualQuality-R1] Score less than 1.0 (Issue #11) CAUTION: External email. Do not reply, click on links or open attachments unless you recognize the sender and know the content is safe. [https://avatars.githubusercontent.com/u/149365261?s=20&v=4]offline888 left a comment (TianheWu/VisualQuality-R1#11)<#11 (comment)> Snipaste_2026-01-21_16-30-36.png (view on web)<https://github.com/user-attachments/assets/6b137135-16f4-430c-b47d-ec018b7916bb> Yes, I've already resized it, but still got a score less than 1.0 ― Reply to this email directly, view it on GitHub<#11 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/APC47QXRRLELY53EW5HU5I34H42WFAVCNFSM6AAAAACSL2MOY6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTONZWHAYDMNJVHA>. You are receiving this because you commented.Message ID: ***@***.***>

[offline888]

I observed the similar phenomenon on AGIQA-3K. I haven't tested other datasets yet.