add qwen2.5 32b as the default

backend/routes/questions.py

@@ -37,7 +37,7 @@ async def get_benchmark_questions(session_id: str):
             if single_dataset and len(single_dataset['train']) > 0:
                 # Prendre 2 questions à partir de l'index 1 (en évitant la première question)
                 start_idx = 1
-                max_questions = min(2, max(0, len(single_dataset['train']) - start_idx))
+                max_questions = min(5, max(0, len(single_dataset['train']) - start_idx))
                 for i in range(max_questions):
                     idx = start_idx + i
                     questions.append({

backend/tasks/create_bench_config_file.py

@@ -122,7 +122,8 @@ class CreateBenchConfigTask:
         # Define required models
         required_models = [
             # "Qwen/Qwen2.5-72B-Instruct"
-            "meta-llama/Llama-3.1-8B-Instruct"
+            # "meta-llama/Llama-3.1-8B-Instruct"
+            "Qwen/Qwen2.5-32B-Instruct"
         ]
         
         # Track found models
@@ -165,11 +166,11 @@ class CreateBenchConfigTask:
             "model_list": model_list,
             
             "model_roles": {
-                "ingestion": ["meta-llama/Llama-3.1-8B-Instruct"],
-                "summarization": ["meta-llama/Llama-3.1-8B-Instruct"],
+                "ingestion": ["Qwen/Qwen2.5-32B-Instruct"],
+                "summarization": ["Qwen/Qwen2.5-32B-Instruct"],
                 "chunking": ["intfloat/multilingual-e5-large-instruct"],
-                "single_shot_question_generation": ["meta-llama/Llama-3.1-8B-Instruct"],
-                "multi_hop_question_generation": ["meta-llama/Llama-3.1-8B-Instruct"],
+                "single_shot_question_generation": ["Qwen/Qwen2.5-32B-Instruct"],
+                "multi_hop_question_generation": ["Qwen/Qwen2.5-32B-Instruct"],
             },
             "pipeline": {
                 "ingestion": {
@@ -200,7 +201,7 @@ class CreateBenchConfigTask:
                     "additional_instructions": "Generate rich and creative questions to test a curious adult",
                     "chunk_sampling": {
                         "mode": "count",
-                        "value": 5,
+                        "value": 10,
                         "random_seed": 123,
                     },
                 },

backend/tasks/evaluation_task.py

@@ -319,6 +319,7 @@ TASKS_TABLE = [yourbench]
         models = [
             "Qwen/QwQ-32B",
             "Qwen/Qwen2.5-72B-Instruct",
+            "Qwen/Qwen2.5-32B-Instruct",
             "meta-llama/Llama-3.1-8B-Instruct",
             "meta-llama/Llama-3.3-70B-Instruct",
             "deepseek-ai/DeepSeek-R1-Distill-Llama-70B",

backend/tasks/get_available_model_provider.py

@@ -136,6 +136,7 @@ if __name__ == "__main__":
         "deepseek-ai/DeepSeek-R1-Distill-Llama-70B",
         "mistralai/Mistral-Small-24B-Instruct-2501",
         "meta-llama/Llama-3.1-8B-Instruct",
+        "Qwen/Qwen2.5-32B-Instruct"
     ]
 
     providers = []