{"author":"qazal","author_email":"77887910+Qazalin@users.noreply.github.com","author_time":1702060210,"commit_time":1702060210,"committer":"GitHub","committer_email":"noreply@github.com","hash":"73b067f5ce00efdffe4a2f5478ac9199c945efff","message":"Bitcast p2 bfloat16 tests + clang fix (#2635)\n\n* add bf16 test support\r\n\r\nthis model takes me almost a minute to download though:\r\n\r\nhttps://huggingface.co/TinyPixel/Llama-2-7B-bf16-sharded/resolve/main/pytorch_model-00001-of-00014.bin?download=true: 100%|█████████████████████████████| 981M/981M [00:40<00:00, 24.2MB/s]\r\n\r\n* ensure we first load if it is bitcast to avoid taking the address of an rvalue\r\n\r\n* tiny bf16 in the cloud\r\n\r\nskip GPU\r\n\r\n* should skip torch\r\n\r\nlint\r\n\r\n* Revert \"ensure we first load if it is bitcast to avoid taking the address of an rvalue\"\r\n\r\nThis reverts commit b86a28ab84bc1173764b2d480218e8de41a32390.\r\n\r\n* break the kernel\r\n\r\n* skip LLVM and GPU in CI\r\n\r\n* skip CUDA","parents":["a29538a094b4c87ae390c33b0c366f2471ff35bd"],"tree_hash":"3893c482aae807d41689cbbb5624bc89011a68bd"}