Remove all traces of low_cpu_mem_usage (#38792)

* remove it from all py files

* remove it from the doc

* remove it from examples

* style

* remove traces of _fast_init

* Update test_peft_integration.py

* CIs

tests/quantization/autoawq/test_awq.py

@@ -328,7 +328,6 @@ class AwqFusedTest(unittest.TestCase):
         model = AutoModelForCausalLM.from_pretrained(
             self.model_name,
             quantization_config=quantization_config,
-            low_cpu_mem_usage=True,
             revision=self.model_revision,
         ).to(torch_device)
 
@@ -347,7 +346,6 @@ class AwqFusedTest(unittest.TestCase):
         model = AutoModelForCausalLM.from_pretrained(
             model_id,
             quantization_config=quantization_config,
-            low_cpu_mem_usage=True,
         ).to(torch_device)
 
         # Check if model has been correctly fused
@@ -370,7 +368,6 @@ class AwqFusedTest(unittest.TestCase):
         model = AutoModelForCausalLM.from_pretrained(
             self.model_name,
             quantization_config=quantization_config,
-            low_cpu_mem_usage=True,
             revision=self.model_revision,
         ).to(torch_device)
 
@@ -399,7 +396,6 @@ class AwqFusedTest(unittest.TestCase):
         model = AutoModelForCausalLM.from_pretrained(
             self.model_name,
             quantization_config=quantization_config,
-            low_cpu_mem_usage=True,
             revision=self.model_revision,
         ).to(torch_device)
 

tests/quantization/hqq/test_hqq.py

@@ -42,7 +42,6 @@ class HQQLLMRunner:
             torch_dtype=compute_dtype,
             device_map=device,
             quantization_config=quant_config,
-            low_cpu_mem_usage=True,
             cache_dir=cache_dir,
         )
         self.tokenizer = AutoTokenizer.from_pretrained(model_id, cache_dir=cache_dir)
@@ -233,7 +232,9 @@ class HQQSerializationTest(unittest.TestCase):
 
         # Load and check if the logits match
         model_loaded = AutoModelForCausalLM.from_pretrained(
-            "quant_model", torch_dtype=torch.float16, device_map=torch_device, low_cpu_mem_usage=True
+            "quant_model",
+            torch_dtype=torch.float16,
+            device_map=torch_device,
         )
 
         with torch.no_grad():