OpenGVLab / Vision-RWKV

Vision-RWKV: Efficient and Scalable Visual Perception with RWKV-Like Architectures
https://arxiv.org/abs/2403.02308
Apache License 2.0
288 stars 11 forks source link

能否做一个替代GPT-4vision能力的版本,我们正在做一个商业&公益兼具的产品 #2

Open baleli668 opened 4 months ago

baleli668 commented 4 months ago

大神,你好 我们正在使用GPT-4vision能力做一款商业&公益兼具的产品,但GPT-4vision价格贵、处理速度慢, 是否可以跟你合作,使用vision-RWKV做出能力替代?

期待你的回信,也可以微信联系我:kobe_heizi

感谢

BlinkDL commented 4 months ago

@baleli668 你可以看 https://github.com/howard-hou/VisualRWKV

baleli668 commented 3 months ago

感谢 pengbo, 有部署好可以直接体验效果的吗?

---原始邮件--- 发件人: "PENG @.> 发送时间: 2024年3月4日(周一) 下午4:43 收件人: @.>; 抄送: @.**@.>; 主题: Re: [OpenGVLab/Vision-RWKV] 能否做一个替代GPT-4vision能力的版本,我们正在做一个商业&公益兼具的产品 (Issue #2)

@baleli668 你可以看 https://github.com/howard-hou/VisualRWKV

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: @.***>

BlinkDL commented 3 months ago

@baleli668 在 https://huggingface.co/spaces/BlinkDL/RWKV-Gradio-1 是 Visual RWKV v5 1.5B

baleli668 commented 3 months ago

bopeng,你好 目前vision-RWKV与gpt-4vision或者通义、阶跃星辰等的图片识别理解差距非常大,暂时无法达到商用状态。

------------------ 原始邮件 ------------------ 发件人: "OpenGVLab/Vision-RWKV" @.>; 发送时间: 2024年3月4日(星期一) 下午4:43 @.>; @.**@.>; 主题: Re: [OpenGVLab/Vision-RWKV] 能否做一个替代GPT-4vision能力的版本,我们正在做一个商业&公益兼具的产品 (Issue #2)

@baleli668 你可以看 https://github.com/howard-hou/VisualRWKV

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: @.***>