[Model] Pass param prefix to LLMHead (#24862)

Signed-off-by: whx-sjtu <2952154980@qq.com>
This commit is contained in:
whx
2025-09-17 16:01:27 +08:00
committed by GitHub
parent 03191cd8f0
commit 4a9375fe9d
58 changed files with 102 additions and 31 deletions
+3
View File
@@ -15,6 +15,8 @@ from vllm.model_executor.layers.vocab_parallel_embedding import (
from vllm.model_executor.model_loader.weight_utils import default_weight_loader
from vllm.model_executor.sampling_metadata import SamplingMetadata
from .utils import maybe_prefix
class ResidualBlock(nn.Module):
@@ -71,6 +73,7 @@ class Medusa(nn.Module):
config.hidden_size,
org_num_embeddings=self.truncated_vocab_size,
padding_size=DEFAULT_VOCAB_PADDING_SIZE,
prefix=maybe_prefix(prefix, "lm_head"),
)
self.lm_heads = [
self.lm_head for _ in range(self.config.num_heads)