github/ColossalAI
1
0
Fork 0
mirror of https://github.com/hpcaitech/ColossalAI.git synced 2025-06-21 05:04:47 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix

Browse Source
This commit is contained in:
wangbluo 2024-09-25 19:02:21 +08:00
parent 91ed32c256
commit 6705dad41b
3 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
colossalai/shardformer/layer/attn.py
View File

@ -501,7 +501,6 @@ class RingAttention(torch.autograd.Function):
v, v,
sp_group, sp_group,
attention_mask_type, attention_mask_type,
tp_group=None,
cu_seqlens=None, cu_seqlens=None,
max_seqlen=None, max_seqlen=None,
valid_indices=None, valid_indices=None,
@ -510,6 +509,7 @@ class RingAttention(torch.autograd.Function):
deterministic=False, deterministic=False,
return_softmax=False, return_softmax=False,
inner_ring_size=None, inner_ring_size=None,
tp_group=None,
**kwargs, **kwargs,
): ):
""" """

2
colossalai/shardformer/modeling/gpt2.py
View File

@ -866,11 +866,11 @@ def get_gpt2_flash_attention_forward(shard_config: Optional[ShardConfig] = None)
key, key,
value, value,
sp_group, sp_group,
tp_group=tp_group,
**attention_mask, **attention_mask,
dropout_p=dropout_p, dropout_p=dropout_p,
scale=scale, scale=scale,
inner_ring_size=shard_config.inner_ring_size, inner_ring_size=shard_config.inner_ring_size,
tp_group=tp_group,
) )
else: else:
attn_output = ColoAttention.attention(query, key, value, **attention_mask, dropout_p=dropout_p, scale=scale) attn_output = ColoAttention.attention(query, key, value, **attention_mask, dropout_p=dropout_p, scale=scale)

2
colossalai/shardformer/modeling/llama.py
View File

@ -571,9 +571,9 @@ def get_llama_flash_attention_forward(shard_config: ShardConfig, sp_mode=None, s
key_states, key_states,
value_states, value_states,
sp_group, sp_group,
tp_group=tp_group,
**attention_mask, **attention_mask,
inner_ring_size=shard_config.inner_ring_size, inner_ring_size=shard_config.inner_ring_size,
tp_group=tp_group,
) )
elif shard_config.enable_flash_attention: elif shard_config.enable_flash_attention: