Why can't I reproduce your results, and at the same time, there is a significant difference in FLOPS?

[oahcang]

[jameslahm]

Thanks for your interest! Could you please provide more details?

[oahcang]

Thanks for your interest! Could you please provide more details?

Under the same conditions, I tested your v10n model to be 8.6 while in Performance it is 6.7, V10s is 24.8

[jameslahm]

Does this issue #13 help?

[obisidianouc]

Thanks for your interest! Could you please provide more details?

Under the same conditions, I tested your v10n model to be 8.6 while in Performance it is 6.7, V10s is 24.8

I find the same issue. The inference time of v10n is about 8, and of v8n is about 6 on A100 (huggingface ZeroGPU)

[jameslahm]

Thanks for your interest! Could you please provide the details of the benchmark and the results?

I find the same issue. The inference time of v10n is about 8, and of v8n is about 6 on A100 (huggingface ZeroGPU)

[jameslahm]

Please feel free to reopen this issue if you have further questions.