fix dtype issue

Signed-off-by: Kyle Sayers <kylesayrs@gmail.com>

Author: kylesayrs

tests/test_compressors/model_compressors/test_model_compressor.py

@@ -393,7 +393,7 @@ def _get_combined_config(s_config, q_config):
     ],
 )
 def test_compress_decompress_model(model_stub, comp_stub, q_format, s_format):
-    model = AutoModelForCausalLM.from_pretrained(model_stub)
+    model = AutoModelForCausalLM.from_pretrained(model_stub, torch_dtype=torch.float32)
     compressor = ModelCompressor.from_pretrained_model(model, s_format, q_format)
 
     # compress model by eagerly compressing state dict
@@ -423,7 +423,7 @@ def test_compress_decompress_model(model_stub, comp_stub, q_format, s_format):
     true_decompressed_model = AutoModelForCausalLM.from_pretrained(
         comp_stub,
         quantization_config=CompressedTensorsConfig(run_compressed=False),
-        torch_dtype=torch.half,
+        torch_dtype=torch.float32,
     )
     true_decompressed = dict(true_decompressed_model.state_dict())
     true_decompressed = {
@@ -441,5 +441,3 @@ def test_compress_decompress_model(model_stub, comp_stub, q_format, s_format):
     for key in decompressed.keys():
         assert torch.allclose(decompressed[key], true_decompressed[key]), f"{key}"
     del true_decompressed
-
-    exit(0)