Upload generate-responses.py with huggingface_hub

generate-responses.py

@@ -47,7 +47,6 @@ from torch import cuda
 from tqdm.auto import tqdm
 from transformers import AutoTokenizer
 from vllm import LLM, SamplingParams
-from vllm.sampling_params import GuidedDecodingParams
 
 # Enable HF Transfer for faster downloads
 os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"
@@ -260,6 +259,17 @@ def main(
     tokenizer = AutoTokenizer.from_pretrained(model_id)
 
     # Create sampling parameters with guided decoding to enforce valid JSON
+    try:
+        from vllm.sampling_params import GuidedDecodingParams
+        guided = GuidedDecodingParams(json_object=True)
+    except ImportError:
+        try:
+            from vllm.guided_decoding import GuidedDecodingParams
+            guided = GuidedDecodingParams(json_object=True)
+        except ImportError:
+            guided = {"json_object": True}
+            logger.info("Using dict-based guided decoding config")
+
     sampling_params = SamplingParams(
         temperature=temperature,
         top_p=top_p,
@@ -267,7 +277,7 @@ def main(
         min_p=min_p,
         max_tokens=max_tokens,
         repetition_penalty=repetition_penalty,
-        guided_decoding=GuidedDecodingParams(json_object=True),
+        guided_decoding=guided,
     )
 
     # Load dataset