[DFLASH] Support grammar-constrained decoding in speculative verify (#30096)

Co-authored-by: hnyls2002 <lsyincs@gmail.com>
2026-07-25 04:36:46 -07:00
parent 17afd8421f
commit d021990bf5
8 changed files with 118 additions and 10 deletions
@@ -2253,7 +2253,9 @@ class Scheduler(
        self._maybe_namespace_elastic_radix_cache(req)

        if self.spec_algorithm.is_dflash_family():
-            error_msg = validate_dflash_request(req, self.enable_overlap)
+            error_msg = validate_dflash_request(
+                req, self.enable_overlap, self.spec_algorithm
+            )
            if error_msg is not None:
                req.set_finish_with_abort(error_msg)
                self.init_req_max_new_tokens(req)