add more helpful comments

tests/llmcompressor/transformers/compression/run_compressed_configs/fp8_dynamic.yaml

@@ -1,4 +1,4 @@
 cadence: "commit"
 test_type: "regression"
-compressed_model_stub: horheynm/TinyLlama_1.1B_Chat_v1.0_FP8_Dynamic_compressed
-uncompressed_model_stub: horheynm/TinyLlama_1.1B_Chat_v1.0_FP8_Dynamic_uncompressed
+compressed_model_stub: nm-testing/TinyLlama_1.1B_Chat_v1.0_FP8_Dynamic_compressed
+uncompressed_model_stub: nm-testing/TinyLlama_1.1B_Chat_v1.0_FP8_Dynamic_uncompressed

tests/llmcompressor/transformers/compression/run_compressed_configs/w4a16.yaml

@@ -1,4 +1,4 @@
 cadence: "commit"
 test_type: "regression"
-compressed_model_stub: horheynm/TinyLlama_1.1B_Chat_v1.0_W4A16_G128_compressed
-uncompressed_model_stub: horheynm/TinyLlama_1.1B_Chat_v1.0_W4A16_G128_uncompressed
+compressed_model_stub: nm-testing/TinyLlama_1.1B_Chat_v1.0_W4A16_G128_compressed
+uncompressed_model_stub: nm-testing/TinyLlama_1.1B_Chat_v1.0_W4A16_G128_uncompressed

tests/llmcompressor/transformers/compression/run_compressed_configs/w8a16.yaml

@@ -0,0 +1,4 @@
+cadence: "commit"
+test_type: "regression"
+compressed_model_stub: nm-testing/TinyLlama_1.1B_Chat_v1.0_W8A16_G128_compressed
+uncompressed_model_stub: nm-testing/TinyLlama_1.1B_Chat_v1.0_W8A16_G128_uncompressed

tests/llmcompressor/transformers/compression/run_compressed_configs/w8a8.yaml

@@ -1,4 +1,4 @@
 cadence: "commit"
 test_type: "regression"
-compressed_model_stub: horheynm/TinyLlama_1.1B_Chat_v1.0_W8A8_Dynamic_Per_Token_compressed
-uncompressed_model_stub: horheynm/TinyLlama_1.1B_Chat_v1.0_W8A8_Dynamic_Per_Token_uncompressed
+compressed_model_stub: nm-testing/TinyLlama_1.1B_Chat_v1.0_W8A8_Dynamic_Per_Token_compressed
+uncompressed_model_stub: nm-testing/TinyLlama_1.1B_Chat_v1.0_W8A8_Dynamic_Per_Token_uncompressed

tests/llmcompressor/transformers/compression/test_run_compressed.py

@@ -18,9 +18,9 @@
 
 @requires_gpu
 @parameterized_class(parse_params(COMPRESSED_LINEAR_CONFIG_DIR))
-class TestUncompressedDecompressed(unittest.TestCase):
+class Test_Decompressed_Linear_Uncompressed_Linear(unittest.TestCase):
     """
-    Uncompressed-decompressed check
+    Uncompressed-Linear-forward decompressed-Linear-foward check
 
     Uncompressed:  Optimized model saved as run_compressed=False, no need to decompress
     Decompressed:  Optimized model saved as run_compressed=True, and decompressed using
@@ -38,18 +38,24 @@ def setUpClass(cls):
         cls.test_dir = tempfile.mkdtemp()
 
         quantization_config = CompressedTensorsConfig(run_compressed=False)
+
+        # Decompressed using HFQuantizer
+        # Linear foward
         cls.decompressed_model = AutoModelForCausalLM.from_pretrained(
             cls.compressed_model_stub,
             torch_dtype="auto",
             device_map="auto",
             quantization_config=quantization_config,
         )
 
+        # Load model as is at the uncompressed state
+        # Linear forward
         cls.uncompressed_model = AutoModelForCausalLM.from_pretrained(
             cls.uncompressed_model_stub,
             torch_dtype=cls.decompressed_model.dtype,
             device_map=cls.decompressed_model.device,
         )
+        breakpoint()
 
         cls.tokenizer = AutoTokenizer.from_pretrained(cls.compressed_model_stub)
 
@@ -94,9 +100,9 @@ def tearDownClass(cls):
 
 @requires_gpu
 @parameterized_class(parse_params(COMPRESSED_LINEAR_CONFIG_DIR))
-class TestCompressedDecompressed(unittest.TestCase):
+class Test_Compressed_CompressedLinear_Decompressed_Linear(unittest.TestCase):
     """
-    Compressed-decompressed check
+    Compressed-CompresesdLinear, Decompressed-Linear check
 
     Compressed:    Optimized model saved as run_compressed=True, no decompression
     Decompressed:  Optimized model saved as run_compressed=True, and decompressed using
@@ -113,13 +119,15 @@ def setUpClass(cls):
         cls.test_dir = tempfile.mkdtemp()
 
         # Should have CompressedLinear modules
+        # Compressed Linear forward
         cls.compressed_model = AutoModelForCausalLM.from_pretrained(
             cls.compressed_model_stub,
             torch_dtype="auto",
             device_map="auto",
         )
 
         # Should just be linear modules
+        # Linear forward
         quantization_config = CompressedTensorsConfig(run_compressed=False)
         cls.decompressed_model = AutoModelForCausalLM.from_pretrained(
             cls.compressed_model_stub,