diff --git a/deepspeed/runtime/config.py b/deepspeed/runtime/config.py index 78e9eaaf1b3f..26c10f3b0e35 100755 --- a/deepspeed/runtime/config.py +++ b/deepspeed/runtime/config.py @@ -205,7 +205,7 @@ def get_sparse_fixed_config(sparsity): sparsity, SPARSE_HORIZONTAL_GLOBAL_ATTENTION, SPARSE_HORIZONTAL_GLOBAL_ATTENTION_DEFAULT) - num_differnt_global_patterns = get_scalar_param( + num_different_global_patterns = get_scalar_param( sparsity, SPARSE_NUM_DIFFERENT_GLOBAL_PATTERNS, SPARSE_NUM_DIFFERENT_GLOBAL_PATTERNS_DEFAULT) @@ -218,7 +218,7 @@ def get_sparse_fixed_config(sparsity): SPARSE_NUM_GLOBAL_BLOCKS: num_global_blocks, SPARSE_ATTENTION_TYPE: attention, SPARSE_HORIZONTAL_GLOBAL_ATTENTION: horizontal_global_attention, - SPARSE_NUM_DIFFERENT_GLOBAL_PATTERNS: num_differnt_global_patterns + SPARSE_NUM_DIFFERENT_GLOBAL_PATTERNS: num_different_global_patterns } diff --git a/deepspeed/runtime/constants.py b/deepspeed/runtime/constants.py index 57bd5ecf81fc..ea767b93e68d 100755 --- a/deepspeed/runtime/constants.py +++ b/deepspeed/runtime/constants.py @@ -40,7 +40,7 @@ SPARSE_ATTENTION_TYPE_DEFAULT = "bidirectional" SPARSE_HORIZONTAL_GLOBAL_ATTENTION = "horizontal_global_attention" SPARSE_HORIZONTAL_GLOBAL_ATTENTION_DEFAULT = False -SPARSE_NUM_DIFFERENT_GLOBAL_PATTERNS = "num_differnt_global_patterns" +SPARSE_NUM_DIFFERENT_GLOBAL_PATTERNS = "num_different_global_patterns" SPARSE_NUM_DIFFERENT_GLOBAL_PATTERNS_DEFAULT = 1 SPARSE_NUM_RANDOM_BLOCKS = "num_random_blocks" SPARSE_NUM_RANDOM_BLOCKS_DEFAULT = 0