Fix AutoModel can't load gptq model due to module prefix mismatch vs AutoModelForCausalLM #49

Job	Run time
build (3.10, macos-15, test_modeling_gemma2.py)	1h 0m 21s
build (3.10, macos-15, test_modeling_gemma.py)	54m 39s
build (3.10, macos-15, test_modeling_llama.py)	11m 35s
build (3.10, macos-15, test_modeling_olmo.py)	12m 25s
build (3.10, macos-15, test_modeling.py)	9m 11s
build (3.10, macos-15, test_modeling_qwen2.py)	4m 8s
build (3.11, macos-15, test_modeling_gemma2.py)	1h 5m 56s
build (3.11, macos-15, test_modeling_gemma.py)	49m 18s
build (3.11, macos-15, test_modeling_llama.py)	11m 28s
build (3.11, macos-15, test_modeling_olmo.py)	9m 52s
build (3.11, macos-15, test_modeling.py)	8m 50s
build (3.11, macos-15, test_modeling_qwen2.py)	4m 23s
build (3.12, macos-15, test_modeling_gemma2.py)	56m 16s
build (3.12, macos-15, test_modeling_gemma.py)	54m 40s
build (3.12, macos-15, test_modeling_llama.py)	14m 18s
build (3.12, macos-15, test_modeling_olmo.py)	7m 41s
build (3.12, macos-15, test_modeling.py)	8m 54s
build (3.12, macos-15, test_modeling_qwen2.py)	4m 55s
	7h 28m 50s

Provide feedback