ci: simplify multi-modality tests by using mixins (#9006)

2025-08-17 13:25:02 +08:00
parent 66d6be0874
commit 1df84ff414
6 changed files with 264 additions and 400 deletions
--- a/python/sglang/srt/multimodal/processors/base_processor.py
+++ b/python/sglang/srt/multimodal/processors/base_processor.py
@@ -217,9 +217,9 @@ class BaseMultimodalProcessor(ABC):
        if videos:
            kwargs["videos"] = videos
        if audios:
-            if self.arch in {
-                "Gemma3nForConditionalGeneration",
-                "Qwen2AudioForConditionalGeneration",
+            if self._processor.__class__.__name__ in {
+                "Gemma3nProcessor",
+                "Qwen2AudioProcessor",
            }:
                # Note(Xinyuan): for gemma3n, ref: https://github.com/huggingface/transformers/blob/ccf2ca162e33f381e454cdb74bf4b41a51ab976d/src/transformers/models/gemma3n/processing_gemma3n.py#L107
                kwargs["audio"] = audios