backend sampling: support returning post-sampling probs (#22622)

* server: Never return 0.0 post-sampling probabilities * backend sampling: support returning post-sampling probs
2026-05-10 19:12:02 +02:00
parent 5d5d2e15d2
commit 2e97c5f96f
4 changed files with 80 additions and 16 deletions
@@ -108,6 +108,7 @@ class ServerProcess:
    no_cache_idle_slots: bool = False
    log_path: str | None = None
    webui_mcp_proxy: bool = False
+    backend_sampling: bool = False
    gcp_compat: bool = False

    # session variables
@@ -252,6 +253,8 @@ class ServerProcess:
            server_args.append("--no-cache-idle-slots")
        if self.webui_mcp_proxy:
            server_args.append("--webui-mcp-proxy")
+        if self.backend_sampling:
+            server_args.append("--backend_sampling")
        if self.gcp_compat:
            env["AIP_MODE"] = "PREDICTION"