正在加载视频...
视频加载失败
the FP8 values in your model after 50 layers of quantize/dequantize operations
10 条评论

tenderizzation1 年前
ty @Norapom04 for finding the clip

adheep1 年前
Wait WTF it ended right

Viv1 年前
the hidden residual stream came through in the end

adheep1 年前
That third kid fucked it up

gslaller1 年前
Is this an illustration of the U-Net architecture?

mobicham1 年前
Waiting for the FP4 meme version 👀

c1 年前
lmfao

Lost(e/employed)1 年前
Isn’t quantization aware training adapted when they release a quantised model?

secemp1 年前
do 1bit next

Satyansh1 年前
chinese to japanese transition was smooth in the middle lmao
