{"author":"George Hotz","author_email":"72895+geohot@users.noreply.github.com","author_time":1700165393,"commit_time":1700165393,"committer":"GitHub","committer_email":"noreply@github.com","hash":"3baaf298d6a1f41e487669cbefbe9f98c2898fdc","message":"two stage cumsum in tensor.py (#2331)\n\n* two stage cumsum in tensor.py\r\n\r\n* 2 more kernels for llama cumsum\r\n\r\n* gpt-2 and llama use fast multinomial","parents":["163b2bc26a1f1451064ac71ac8bc4548da73cd86"],"tree_hash":"2690e24b56682ffbf701afe8c30b4f58034f3ec7"}