Revert "[Bugfix] Limit profiling run sequence length by max_model_len (#14785) (#14892)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
2025-03-17 00:13:46 +08:00
parent e53b1350f2
commit f6137adbcb
5 changed files with 0 additions and 9 deletions
--- a/vllm/inputs/registry.py
+++ b/vllm/inputs/registry.py
@@ -330,11 +330,6 @@ class InputRegistry:
        from vllm.multimodal import MultiModalKwargs
        from vllm.multimodal.profiling import MultiModalProfiler

-        if seq_len > model_config.max_model_len:
-            raise AssertionError(
-                f"Profiling attempted with sequence length ({seq_len}) "
-                f"greater than model length ({model_config.max_model_len})")
-
        if mm_registry.has_processor(model_config):
            tokenizer = cached_tokenizer_from_config(model_config)
            processor = mm_registry.create_processor(model_config,