model_free_ptq

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

Author: kylesayrs

src/llmcompressor/__init__.py

@@ -26,4 +26,4 @@
     create_session,
     reset_session,
 )
-from llmcompressor.entrypoints import Oneshot, oneshot, train, ptq_weights
+from llmcompressor.entrypoints import Oneshot, oneshot, train, model_free_ptq

src/llmcompressor/entrypoints/__init__.py

@@ -9,5 +9,5 @@
 
 from .oneshot import Oneshot, oneshot
 from .train import train
-from .weights_ptq import ptq_weights
+from .model_free import model_free_ptq
 from .utils import post_process, pre_process

…ssor/entrypoints/weights_ptq/__init__.py …essor/entrypoints/model_free/__init__.pysrc/llmcompressor/entrypoints/weights_ptq/__init__.py renamed to src/llmcompressor/entrypoints/model_free/__init__.py src/llmcompressor/entrypoints/weights_ptq/__init__.py renamed to src/llmcompressor/entrypoints/model_free/__init__.py

@@ -11,28 +11,28 @@
 from loguru import logger
 from safetensors.torch import load_file, save_file
 
-from llmcompressor.entrypoints.weights_ptq.helpers import (
+from llmcompressor.entrypoints.model_free.helpers import (
     gpu_if_available,
     validate_scheme,
 )
-from llmcompressor.entrypoints.weights_ptq.lifecycle import (
+from llmcompressor.entrypoints.model_free.lifecycle import (
     calibrate_weights,
     compress_module,
     initialize_quantized_linear,
 )
-from llmcompressor.entrypoints.weights_ptq.model_utils import (
+from llmcompressor.entrypoints.model_free.model_utils import (
     get_checkpoint_files,
     is_weights_file,
 )
-from llmcompressor.entrypoints.weights_ptq.save_utils import (
+from llmcompressor.entrypoints.model_free.save_utils import (
     update_config,
     update_safetensors_index,
 )
 
-__all__ = ["ptq_weights"]
+__all__ = ["model_free_ptq"]
 
 
-def ptq_weights(
+def model_free_ptq(
     model_stub: str | os.PathLike,
     save_directory: str | os.PathLike,
     scheme: QuantizationScheme | str,
@@ -41,7 +41,8 @@ def ptq_weights(
     device: Optional[torch.device | str] = None,
 ):
     """
-    Name TBD
+    Quantize a model without the need for a model definition. This function operates on
+    a model stub or folder containing weights saved in safetensors files
 
     :param model_stub: huggingface model hub or path to local weights files
     :param scheme: weight quantization scheme or preset scheme name

…essor/entrypoints/weights_ptq/helpers.py …ressor/entrypoints/model_free/helpers.pysrc/llmcompressor/entrypoints/weights_ptq/helpers.py renamed to src/llmcompressor/entrypoints/model_free/helpers.py src/llmcompressor/entrypoints/weights_ptq/helpers.py renamed to src/llmcompressor/entrypoints/model_free/helpers.py

@@ -6,7 +6,7 @@
 from compressed_tensors.quantization import QuantizationArgs, QuantizationScheme
 from safetensors.torch import load_file
 
-from llmcompressor import oneshot, ptq_weights
+from llmcompressor import model_free_ptq, oneshot
 from llmcompressor.modifiers.quantization import QuantizationModifier
 from tests.testing_utils import requires_gpu
 
@@ -51,7 +51,7 @@ def test_weights_ptq_e2e(scheme, tmp_path):
     ptq_outdir = tmp_path / "weights_out"
     oneshot_outdir = tmp_path / "oneshot_out"
 
-    ptq_weights(
+    model_free_ptq(
         model,
         ptq_outdir,
         scheme=scheme,