5 Commits

Author	SHA1	Message	Date
戒酒的李白	3efea929c8	The multi-head attention mechanism is basically completed.	2024-10-13 10:04:18 +08:00
戒酒的李白	9af61e2ade	Calculates the scaling dot product attention	2024-10-07 09:51:29 +08:00
戒酒的李白	4500b2719e	Divide the input into long heads	2024-10-06 11:54:32 +08:00
戒酒的李白	f5e307d3f8	Define the linear transformation layer	2024-10-06 11:34:31 +08:00
戒酒的李白	ee739c3c81	Multi-head attention mechanism infrastructure and input dimension settings.	2024-10-05 00:49:24 +08:00