Skip to content

PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度

Fix shard finalization issue to prevent skipping non-LLM tail layers (#1548)

Liang Lv81e79a14da55
diff --git a/auto_round/compressors/shard_writer.py b/auto_round/compressors/shard_writer.py
index b6dd329e3..20e959e8e 100644
--- a/auto_round/compressors/shard_writer.py
+++ b/auto_round/compressors/shard_writer.py
@@ -170,7 +170,9 @@ class ShardWriter:
"""Saves remaining weights, renames files, and writes the index JSON."""
# 1. Capture remaining weights not yet saved
full_sd = self.model.state_dict()
- tie_word_embeddings = getattr(getattr(self.model, "config", None), "tie_word_embeddings", True)
+ tie_word_embeddings = False
+ if hasattr(self.model, "config") and hasattr(self.model.config, "tie_word_embeddings"):
+ tie_word_embeddings = self.model.config.tie_word_embeddings
finalize_skipped_meta_tensors = []
for pname, tensor in full_sd.items():
diff --git a/test/test_cpu/utils/test_shard_writer.py b/test/test_cpu/utils/test_shard_writer.py
new file mode 100644
index 000000000..2dc7c6b72
--- /dev/null
+++ b/test/test_cpu/utils/test_shard_writer.py
@@ -0,0 +1,96 @@
+# Copyright (c) 2026 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import os
+from types import SimpleNamespace
+
+import torch
+
+from auto_round.compressors.shard_writer import ShardWriter
+
+
+class _ToyBlock(torch.nn.Module):
+ def __init__(self):
+ super().__init__()
+ self.linear = torch.nn.Linear(4, 4)
+
+
+class _DiffusionStyleModel(torch.nn.Module):
+ def __init__(self):
+ super().__init__()
+ self.transformer_blocks = torch.nn.ModuleList([_ToyBlock()])
+ self.proj_out = torch.nn.Linear(4, 2)
+ self.config = SimpleNamespace(model_type="toy-diffusion")
+
+
+class _RounderStub:
+ def __init__(self, model, output_dir):
+ self.model = model
+ self.bits = 4
+ self.formats = [object()]
+ self.max_shard_size = "1MB"
+ self.safe_serialization = False
+ self._output_dir = output_dir
+
+ def _get_save_folder_name(self, _format):
+ return self._output_dir
+
+
+def test_finalize_saves_tail_layer_when_tie_word_embeddings_missing(tmp_path):
+ model = _DiffusionStyleModel()
+ rounder = _RounderStub(model, str(tmp_path))
+ writer = ShardWriter(rounder)
+
+ assert writer.lm_head_name == "proj_out"
+ assert not hasattr(model.config, "tie_word_embeddings")
+
+ writer.save_module(model.transformer_blocks[0], "transformer_blocks.0")
+ writer.finalize()
+
+ shard_path = os.path.join(tmp_path, "model.bin")
+ saved_tensors = torch.load(shard_path, map_location="cpu")
+
+ assert "transformer_blocks.0.linear.weight" in saved_tensors
+ assert "proj_out.weight" in saved_tensors, "proj_out must be saved when tie_word_embeddings is absent"
+ assert "proj_out.bias" in saved_tensors
+
+
+class _LMStyleModel(torch.nn.Module):
+ """Model whose config explicitly sets tie_word_embeddings=True."""
+
+ def __init__(self):
+ super().__init__()
+ self.transformer_blocks = torch.nn.ModuleList([_ToyBlock()])
+ self.lm_head = torch.nn.Linear(4, 2, bias=False)
+ self.config = SimpleNamespace(model_type="toy-lm", tie_word_embeddings=True)
+
+
+def test_finalize_skips_lm_head_when_tie_word_embeddings_true(tmp_path):
+ """Complementary test: when tie_word_embeddings=True the lm_head should be
+ skipped (not written to disk) and offloaded to meta."""
+ model = _LMStyleModel()
+ rounder = _RounderStub(model, str(tmp_path))
+ writer = ShardWriter(rounder)
+
+ assert writer.lm_head_name == "lm_head"
+
+ writer.save_module(model.transformer_blocks[0], "transformer_blocks.0")
+ writer.finalize()
+
+ shard_path = os.path.join(tmp_path, "model.bin")
+ saved_tensors = torch.load(shard_path, map_location="cpu")
+
+ assert "transformer_blocks.0.linear.weight" in saved_tensors
+ assert "lm_head.weight" not in saved_tensors, "lm_head must be skipped when tied"
+ assert model.lm_head.weight.device.type == "meta"