[Clean-up] Remove SparseAutoModelForCausalLM and sparsification folder (#1826)

Summary
- Kill folder
- Move compressed_tensors_utils to compression folder

Good riddance

Author: dsikka

.github/workflows/test-check-transformers.yaml

@@ -101,10 +101,6 @@ jobs:
         if: (success() || failure()) && steps.install.outcome == 'success'
         run: |
           pytest -v tests/llmcompressor/transformers/oneshot
-      - name: Running Sparsification Tests
-        if: (success() || failure()) && steps.install.outcome == 'success'
-        run: |
-          pytest tests/llmcompressor/transformers/sparsification -v
       - name: Running OBCQ Tests
         if: (success() || failure()) && steps.install.outcome == 'success'
         run: |

docs/guides/saving_a_model.md

@@ -69,7 +69,7 @@ If you need more control, you can wrap `save_pretrained` manually:
 
 ```python
 from transformers import AutoModelForCausalLM
-from llmcompressor.transformers.sparsification.compressed_tensors_utils import modify_save_pretrained
+from llmcompressor.transformers.compression.compressed_tensors_utils import modify_save_pretrained
 
 # Load model
 model = AutoModelForCausalLM.from_pretrained("your-model")

src/llmcompressor/entrypoints/utils.py

@@ -23,7 +23,7 @@
 )
 from llmcompressor.core import reset_session
 from llmcompressor.pytorch.model_load.helpers import parse_dtype
-from llmcompressor.transformers.sparsification.compressed_tensors_utils import (
+from llmcompressor.transformers.compression.compressed_tensors_utils import (
     modify_save_pretrained,
     untie_word_embeddings,
 )

src/llmcompressor/pytorch/model_load/helpers.py

@@ -40,7 +40,7 @@ def save_checkpoint(
     :param save_safetensors: save model checkpoint using safetensors file type
     :param save_compressed: save model checkpoint using compressed-tensors format
     """
-    from llmcompressor.transformers.sparsification.compressed_tensors_utils import (
+    from llmcompressor.transformers.compression.compressed_tensors_utils import (
         get_model_compressor,  # avoid circular import
     )
 

src/llmcompressor/transformers/__init__.py

@@ -6,8 +6,4 @@
 
 # (import order matters for circular import avoidance)
 from .utils import *
-
-from .sparsification import (
-    SparseAutoModelForCausalLM,
-)
 from .finetune import *

src/llmcompressor/transformers/sparsification/compressed_tensors_utils.py renamed to src/llmcompressor/transformers/compression/compressed_tensors_utils.py

@@ -6,7 +6,7 @@
 
 from llmcompressor.core import State
 from llmcompressor.modifiers.transform import QuIPModifier, SpinQuantModifier
-from llmcompressor.transformers.sparsification.compressed_tensors_utils import (
+from llmcompressor.transformers.compression.compressed_tensors_utils import (
     untie_word_embeddings,
 )
 from tests.testing_utils import requires_gpu

src/llmcompressor/transformers/sparsification/__init__.py

@@ -22,14 +22,14 @@
 from llmcompressor import oneshot
 from llmcompressor.core import reset_session
 from llmcompressor.pytorch.utils.helpers import tensor_sparsity
-from llmcompressor.transformers.compression.sparsity_metadata_config import (
-    SparsityConfigMetadata,
-)
-from llmcompressor.transformers.sparsification.compressed_tensors_utils import (
+from llmcompressor.transformers.compression.compressed_tensors_utils import (
     get_model_compressor,
     modify_save_pretrained,
     untie_word_embeddings,
 )
+from llmcompressor.transformers.compression.sparsity_metadata_config import (
+    SparsityConfigMetadata,
+)
 from tests.testing_utils import requires_gpu