biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[V0 Deprecation] Remove V0 sampling metadata (#25345)

Browse Source 
Signed-off-by: Woosuk Kwon <woosuk@thinkingmachines.ai>
This commit is contained in:
Woosuk Kwon
2025-09-21 10:37:11 -07:00
committed by GitHub
parent c438b2951c
commit 1c3ffdbecc
141 changed files with 172 additions and 583 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
vllm/model_executor/models/granite.py
View File

@@ -48,7 +48,6 @@ from vllm.model_executor.layers.vocab_parallel_embedding import (
DEFAULT_VOCAB_PADDING_SIZE, ParallelLMHead, VocabParallelEmbedding)
from vllm.model_executor.model_loader.weight_utils import (
default_weight_loader, maybe_remap_kv_scale_name)
from vllm.model_executor.sampling_metadata import SamplingMetadata
from vllm.sequence import IntermediateTensors
from .interfaces import SupportsLoRA, SupportsPP
@@ -463,11 +462,9 @@ class GraniteForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
inputs_embeds)
return model_output
def compute_logits(
self, hidden_states: torch.Tensor,
sampling_metadata: SamplingMetadata) -> Optional[torch.Tensor]:
logits = self.logits_processor(self.lm_head, hidden_states,
sampling_metadata)
def compute_logits(self,
hidden_states: torch.Tensor) -> Optional[torch.Tensor]:
logits = self.logits_processor(self.lm_head, hidden_states)
return logits
def make_empty_intermediate_tensors(