13 lines
544 B
Python
13 lines
544 B
Python
|
from vllm.v1.worker.gpu import input_batch
|
||
|
|
from vllm.v1.worker.gpu.sample import gumbel, logprob, penalties
|
||
|
|
|
||
|
|
from vllm_ascend.worker.v2.input_batch import post_update
|
||
|
|
from vllm_ascend.worker.v2.sample.gumbel import gumbel_sample
|
||
|
|
from vllm_ascend.worker.v2.sample.logprob import compute_token_logprobs
|
||
|
|
from vllm_ascend.worker.v2.sample.penalties import apply_penalties
|
||
|
|
|
||
|
|
logprob.compute_token_logprobs = compute_token_logprobs
|
||
|
|
penalties.apply_penalties = apply_penalties
|
||
|
|
gumbel.gumbel_sample = gumbel_sample
|
||
|
|
input_batch.post_update = post_update
|