xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-31 12:47:10 +08:00
Code Issues Packages Projects Releases Wiki Activity

compute_logits

Browse Source 
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
This commit is contained in:
Woosuk Kwon 2025-09-21 11:26:33 -07:00
parent 631b5b47c1
commit bc73f674bb
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
vllm/v1/worker/gpu/model_runner.py
View File

@ -178,7 +178,7 @@ class GPUModelRunner:
num_reqs=num_reqs, num_reqs=num_reqs,
device=self.device, device=self.device,
) )
logits = self.model.compute_logits(hidden_states, None) logits = self.model.compute_logits(hidden_states)
self.sampler(logits, sampling_metadata) self.sampler(logits, sampling_metadata)
def profile_run(self) -> None: def profile_run(self) -> None:
@ -377,7 +377,7 @@ class GPUModelRunner:
input_batch: InputBatch, input_batch: InputBatch,
) -> SamplerOutput: ) -> SamplerOutput:
sample_hidden_states = hidden_states[input_batch.logits_indices] sample_hidden_states = hidden_states[input_batch.logits_indices]
logits = self.model.compute_logits(sample_hidden_states, None) logits = self.model.compute_logits(sample_hidden_states)
pos = input_batch.positions[input_batch.logits_indices] pos = input_batch.positions[input_batch.logits_indices]
idx_mapping_np = input_batch.idx_mapping_np idx_mapping_np = input_batch.idx_mapping_np
num_reqs = logits.shape[0] num_reqs = logits.shape[0]