:wrench: TEST no quantisation

moral_compass_demo.py

@@ -16,7 +16,7 @@ qwen_model = AutoModelForCausalLM.from_pretrained(
     qwen_model_name,
     torch_dtype=torch.bfloat16, #testing for underflow issues
     device_map="auto",
-    quantization_config = quantization_config,
+    # quantization_config = quantization_config,
 )
 
 qwen_generationconfig = GenerationConfig(
@@ -58,7 +58,7 @@ llama2_pipe = pipeline(
     model=llama2_model_id,
     torch_dtype=torch.bfloat16,
     device_map="auto",
-    model_kwargs={"quantization_config": quantization_config},
+    # model_kwargs={"quantization_config": quantization_config},
     # quantization_config=quantization_config,
 )
 
@@ -69,7 +69,7 @@ llama3_pipe = pipeline(
     model=llama3_model_id,
     torch_dtype=torch.bfloat16,
     device_map="auto",
-    model_kwargs={"quantization_config": quantization_config},
+    # model_kwargs={"quantization_config": quantization_config},
 )
 
 @spaces.GPU