chenchenghao/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 11 Packages Projects Releases Wiki Activity

Minor fix in compiler & format (#545)

Browse Source
This commit is contained in:
sglang
2024-06-29 23:42:14 -07:00
committed by GitHub
parent 9ce89bc14b
commit 11616fc6bd
12 changed files with 28 additions and 33 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
python/sglang/srt/models/chatglm.py
View File

@@ -36,7 +36,6 @@ LoraConfig = None
class GLMAttention(nn.Module):
def __init__(
self,
config,
@@ -294,7 +293,6 @@ class GLMTransformer(nn.Module):
class ChatGLMModel(nn.Module):
def __init__(
self,
config,

1
python/sglang/srt/models/grok.py
View File

@@ -521,7 +521,6 @@ class Grok1DecoderLayer(nn.Module):
hidden_states: torch.Tensor,
input_metadata: InputMetadata,
) -> torch.Tensor:
hidden_states = (
self.post_attn_norm(
self.self_attn(

6
python/sglang/srt/models/llama2.py
View File

@@ -160,9 +160,9 @@ class LlamaDecoderLayer(nn.Module):
if rope_scaling is not None and getattr(
config, "original_max_position_embeddings", None
):
rope_scaling["original_max_position_embeddings"] = (
config.original_max_position_embeddings
)
rope_scaling[
"original_max_position_embeddings"
] = config.original_max_position_embeddings
max_position_embeddings = getattr(config, "max_position_embeddings", 8192)
self.self_attn = LlamaAttention(
hidden_size=self.hidden_size,