transformer-shortest-paths - Experimentally evaluating transformer's generalization on a synthetic task

Age	Commit message (Collapse)	Author
12 days	addHEAD main	SIPB

12 days	img	SIPB

12 days	add stuff	SIPB

12 days	Commit everything	SIPB

12 days	adding some stuff	SIPB

14 days	Add tune loss too	SIPB

14 days	Add nearly finished code	SIPB

14 days	Commit more stuff	SIPB

14 days	Commit everything	SIPB

2024-12-08	Oh oops delete those files too	SIPB

2024-12-08	Move *.dot to img too	SIPB

2024-12-08	More work on blog.md, move images to img folder	SIPB

2024-12-07	Add super good code	SIPB

2024-12-04	update	SIPB

2024-12-03	Latest blog post and graphs	SIPB

2024-12-02	Latest copy of blog post and insane TSP	SIPB

2024-11-27	Get train err down to .35, add model and loss file	SIPB

2024-11-24	Add first draft of blog post	SIPB

2024-11-21	Make the target embedding learnable	SIPB

2024-11-21	Better plots	SIPB

2024-11-21	New embeddings and readout	SIPB

2024-11-19	Decrease num layers	SIPB

2024-11-17	MORE LAYERS	SIPB

2024-11-17	Use bfloat16	SIPB

2024-11-17	idk what this does but it's the latest version of TSP	SIPB

2024-11-17	Generate data on the fly	SIPB

2024-11-16	Minor README tweaks	SIPB

2024-11-13	updates	Alek Westover

2024-11-01	Delete unused images dir	SIPB

2024-10-28	Condense proposal into 1 page	sipb

2024-10-27	wrote project proposal @anthowan Anthony can you make sure it looks good?	Alek Westover

2024-10-23	Tasks 1 through 4	SIPB

2024-10-23	update	Alek Westover

2024-10-03	fixed some bugs	Alek Westover

2024-10-03	update	Alek Westover

2024-10-03	add readme	Alek Westover

2024-10-03	upload	Alek Westover

2024-10-03	add	Alek Westover