SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

fix typos in the code comments and error messages (#36993)

Browse Source 
* chore: enhance code comments

* chore: enhance code comments

* chore: enhance code comments

* chore: enhance code comments

* chore: enhance code comments

* chore: enhance code comments

* chore: enhance code comments
This commit is contained in:
Afanti
2025-03-27 00:09:48 +08:00
committed by GitHub
parent 79d6f9fd70
commit 44715225e3
25 changed files with 32 additions and 32 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
src/transformers/models/aria/modeling_aria.py
View File

@@ -267,7 +267,7 @@ def sequential_experts_gemm(token_states, expert_weights, tokens_per_expert):
output = torch.zeros(num_tokens, out_features, dtype=token_states.dtype, device=token_states.device)
cumsum_num_tokens = torch.cumsum(tokens_per_expert, dim=0)
# Insert zero at the begining for offset index's convenience
# Insert zero at the beginning for offset index's convenience
zero_tensor = torch.zeros(1, dtype=torch.long, device=cumsum_num_tokens.device)
cumsum_num_tokens = torch.cat((zero_tensor, cumsum_num_tokens))

2
src/transformers/models/aria/modular_aria.py
View File

@@ -86,7 +86,7 @@ def sequential_experts_gemm(token_states, expert_weights, tokens_per_expert):
output = torch.zeros(num_tokens, out_features, dtype=token_states.dtype, device=token_states.device)
cumsum_num_tokens = torch.cumsum(tokens_per_expert, dim=0)
# Insert zero at the begining for offset index's convenience
# Insert zero at the beginning for offset index's convenience
zero_tensor = torch.zeros(1, dtype=torch.long, device=cumsum_num_tokens.device)
cumsum_num_tokens = torch.cat((zero_tensor, cumsum_num_tokens))