File tree Expand file tree Collapse file tree 1 file changed +4
-4
lines changed
examples/06_bmg_flash_attention Expand file tree Collapse file tree 1 file changed +4
-4
lines changed Original file line number Diff line number Diff line change 3838 See https://arxiv.org/pdf/2307.08691 for details of Flash Attention V2 algorithm
3939
4040 To run this example:
41- $ ./examples/sycl/06_bmg_flash_attention_cachedKV/06_bmg_prefill_attention_cachedKV --seq_len_qo=512
42- --seq_len_kv=512 --seq_len_kv_cache=512 --head_size_vo=128 --head_size_qk=128
41+ $ ./examples/06_bmg_flash_attention/06_bmg_chunk_prefill_hdim64 --seq_len_qo=512
42+ --seq_len_kv=512 --seq_len_kv_cache=512 --head_size_vo=64 --head_size_qk=64
4343
4444 Causal masking of the first matrix multiplication is supported (`--is_causal`)
4545
4646 To build & run this example (from your build dir):
4747
48- $ ninja 06_bmg_prefill_attention_cachedKV
49- $ ./examples/sycl/06_bmg_flash_attention_cachedKV/06_bmg_prefill_attention_cachedKV
48+ $ ninja 06_bmg_chunk_prefill_hdim64
49+ $ ./examples/06_bmg_flash_attention/06_bmg_chunk_prefill_hdim64
5050
5151 Call with `--help` for information about available options
5252*/
You can’t perform that action at this time.
0 commit comments