github/ColossalAI
1
0
Fork 0
mirror of https://github.com/hpcaitech/ColossalAI.git synced 2025-09-27 12:43:02 +00:00
Code Issues Packages Projects Releases Wiki Activity

[shardformer] vit/llama/t5 ignore the sequence parallelism flag and some fix. (#4498)

Browse Source 
* [shardformer] chatglm support sequence parallel

[shardformer] chatglm support sequence parallel

[shardformer] chatglm support sequence parallel

[shardformer] chatglm support sequence parallel

[shardformer] chatglm support sequence parallel

[shardformer] chatglm support sequence parallel

* fix

fix

fix

fix

* [shardformer] jit fused fix

* [shardformer] jit fused fix

* [shardformer] jit fused fix

* [shardformer] jit fused fix

* [shardformer] jit fused fix

* [shardformer] jit fused fix

* [shardformer] jit fused fix

* activate checks
This commit is contained in:
flybird11111
2023-08-24 15:50:02 +08:00
committed by GitHub
parent e04436a82a
commit 3353e55c80
7 changed files with 46 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
colossalai/shardformer/policies/t5.py
View File

@@ -1,3 +1,4 @@
import warnings
from functools import partial
from typing import Callable, Dict, List, Optional, Tuple
@@ -59,6 +60,10 @@ class T5BasePolicy(Policy):
policy = {}
if self.shard_config.enable_sequence_parallelism:
self.shard_config.enable_sequence_parallelism = False
warnings.warn("T5 dosen't support sequence parallelism now, will ignore the sequence parallelism flag.")
if self.shard_config.enable_tensor_parallelism:
policy[T5Stack] = ModulePolicyDescription(sub_module_replacement=[
SubModuleReplacementDescription(