[Feature] Online Chat API Support Return logprobs (#2777)

* online chat support logprobs * check xpu * check vl_gpu_model_runner and xpu_model_runner * get_worker() check platform
2025-10-05 16:48:03 +08:00 · 2025-07-10 16:33:40 +08:00
parent 24f934f1f9
commit d33105baeb
22 changed files with 608 additions and 114 deletions
--- a/fastdeploy/model_executor/layers/sample/meta_data.py
+++ b/fastdeploy/model_executor/layers/sample/meta_data.py
@@ -42,3 +42,4 @@ class SamplingMetadata:

    top_p: paddle.Tensor
    top_k: Optional[paddle.Tensor] = None
+    max_num_logprobs: Optional[int] = None