vllm/v1/worker/tpu_model_runner.py · 86cbd2eee97a98df59c531c34d2aeff5a2b5765d · OpenDAS / vllm_cscc · GitLab

Find file Blame History Permalink

[V1][TPU] Do not compile sampling more than needed (#15883) · bd7599d3
Nicolò Lucchesi authored Apr 03, 2025
```
Signed-off-by: NickLucche <nlucches@redhat.com>
```
bd7599d3

tpu_model_runner.py 46.1 KB