{"author":"Arseny Kapoulkine","author_email":"arseny.kapoulkine@gmail.com","author_time":1711406465,"commit_time":1711406465,"committer":"GitHub","committer_email":"noreply@github.com","hash":"cb6e7b57a639a926ebd4df7ff46e6afbb3217009","message":"examples: Fix parameter bandwidth accounting for quantized LLama (#3930)\n\nInstead of assuming every parameter is 2 bytes, just add up tensor sizes\r\nin bytes","parents":["4ecd5789ab4d4f403c00172ff6980e798585beba"],"tree_hash":"8d955d073ba2bd7999f4da9e408a2e762e3776e8"}