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

7
vllm/model_executor/models/molmo.py
View File

@@ -26,7 +26,6 @@ from vllm.distributed import (get_pp_group, get_tensor_model_parallel_rank,
get_tensor_model_parallel_world_size,
split_tensor_along_last_dim,
tensor_model_parallel_all_gather)
from vllm.model_executor import SamplingMetadata
from vllm.model_executor.layers.activation import (MulAndSilu, QuickGELU,
SiluAndMul)
from vllm.model_executor.layers.layernorm import RMSNorm
@@ -1527,10 +1526,8 @@ class MolmoForCausalLM(nn.Module, SupportsMultiModal, SupportsPP, SupportsLoRA,
return hidden_states
def compute_logits(self, hidden_states: torch.Tensor,
sampling_metadata: SamplingMetadata) -> torch.Tensor:
logits = self.logits_processor(self.lm_head, hidden_states,
sampling_metadata)
def compute_logits(self, hidden_states: torch.Tensor) -> torch.Tensor:
logits = self.logits_processor(self.lm_head, hidden_states)
return logits
def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]):