{"author":"chenyu","author_email":"chenyu@fastmail.com","author_time":1699646822,"commit_time":1699646822,"committer":"GitHub","committer_email":"noreply@github.com","hash":"a753c8e07150bb7dfeb83efc801ea2706316f50e","message":"examples of new GPT2 and JIT change (#2261)\n\n* var_vals are global\r\n\r\n* working with global ish\r\n\r\n* better\r\n\r\n* fix export model\r\n\r\n* fix tests\r\n\r\n* better kv cache\r\n\r\n* does it run?\r\n\r\n* use where for kvmask\r\n\r\n* fix excessive var_vals\r\n\r\n* fix import\r\n\r\n* how does multigpu use this?\r\n\r\n* llama kinda work\r\n\r\n* faster and simpler\r\n\r\n* cleanup\r\n\r\n* fix conversation mode\r\n\r\n* test cleanups\r\n\r\n* fix one more test\r\n\r\n* test cleanup\r\n\r\n---------\r\n\r\nCo-authored-by: George Hotz <geohot@gmail.com>","parents":["b6aaf12df7015c7e7d1e292c969e9ea316293224"],"tree_hash":"a278d80da9b22667e49ec8494cdb7d0068546f0e"}