Skip to content

Make OPT policy backward compatible with pre-OPT transformers versions #2254

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
jeffra merged 1 commit into from
Aug 23, 2022
+3 −4
Merged

Make OPT policy backward compatible with pre-OPT transformers versions #2254

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 3 additions & 4 deletions deepspeed/module_inject/replace_policy.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -444,13 +444,12 @@ def __init__(self, client_module, inference=True):
try:
import transformers
HFOPTLayerPolicy._orig_layer_class = transformers.models.opt.modeling_opt.OPTDecoderLayer
if isinstance(DSPolicy.hf_model_config,
transformers.models.opt.configuration_opt.OPTConfig):
self.pre_attn_norm = self.hf_model_config.do_layer_norm_before
except:
HFOPTLayerPolicy._orig_layer_class = None

if isinstance(DSPolicy.hf_model_config,
transformers.models.opt.configuration_opt.OPTConfig):
self.pre_attn_norm = self.hf_model_config.do_layer_norm_before

def get_hidden_heads(self):
return self.client_module.self_attn.embed_dim, \
self.client_module.self_attn.num_heads
Expand Down