Enable Eagle3 speculative decoding for GPT-OSS model (#25246)

Signed-off-by: Eldar Kurtic <8884008+eldarkurtic@users.noreply.github.com>
2025-09-22 10:50:39 +02:00
parent f92d952632
commit 21467f9a1c
3 changed files with 41 additions and 12 deletions
--- a/vllm/config/speculative.py
+++ b/vllm/config/speculative.py
@@ -527,7 +527,7 @@ class SpeculativeConfig:
                             "speculative decoding is > 1, but got "
                             f"{self.disable_by_batch_size=}")

-        eagle3_target_supported = ["llama", "qwen"]
+        eagle3_target_supported = ["llama", "qwen", "gpt_oss"]
        if self.method == "eagle3" and self.target_model_config and not any(
                supported_model in
                self.target_model_config.hf_text_config.model_type