[Cleanup] Remove obsolete spec decoding compatibility logic (#32003)
Signed-off-by: Nick Hill <nickhill123@gmail.com>
This commit is contained in:
@@ -22,7 +22,6 @@ from vllm.v1.sample.logits_processor import (
|
||||
MoveDirectionality,
|
||||
)
|
||||
from vllm.v1.sample.metadata import SamplingMetadata
|
||||
from vllm.v1.spec_decode.utils import is_spec_decode_unsupported
|
||||
from vllm.v1.utils import copy_slice
|
||||
from vllm.v1.worker.block_table import MultiGroupBlockTable
|
||||
|
||||
@@ -176,9 +175,6 @@ class InputBatch:
|
||||
self.top_k_cpu = self.top_k_cpu_tensor.numpy()
|
||||
self.top_k_reqs: set[str] = set()
|
||||
|
||||
# IDs of requests which do not support spec decoding
|
||||
self.spec_decode_unsupported_reqs: set[str] = set()
|
||||
|
||||
# Frequency penalty related data structures
|
||||
self.frequency_penalties = torch.empty(
|
||||
(max_num_reqs,), dtype=torch.float, device=device
|
||||
@@ -346,8 +342,6 @@ class InputBatch:
|
||||
self.block_table.add_row(request.block_ids, req_index)
|
||||
|
||||
if sampling_params := request.sampling_params:
|
||||
if self.is_spec_decode and is_spec_decode_unsupported(sampling_params):
|
||||
self.spec_decode_unsupported_reqs.add(req_id)
|
||||
if sampling_params.sampling_type == SamplingType.GREEDY:
|
||||
# Should avoid division by zero later when apply_temperature.
|
||||
self.temperature_cpu[req_index] = 0.0
|
||||
@@ -510,7 +504,6 @@ class InputBatch:
|
||||
self.random_reqs.discard(req_id)
|
||||
self.top_p_reqs.discard(req_id)
|
||||
self.top_k_reqs.discard(req_id)
|
||||
self.spec_decode_unsupported_reqs.discard(req_id)
|
||||
self.frequency_penalties_reqs.discard(req_id)
|
||||
self.presence_penalties_reqs.discard(req_id)
|
||||
self.repetition_penalties_reqs.discard(req_id)
|
||||
|
||||
Reference in New Issue
Block a user