Defuser 0.0.1__tar.gz → 0.0.2__tar.gz

{defuser-0.0.1 → defuser-0.0.2}/Defuser.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: Defuser
-Version: 0.0.1
+Version: 0.0.2
 Summary: Model defuser helper for HF Transformers.
 Author-email: ModelCloud <qubitium@modelcloud.ai>
 License-Expression: Apache-2.0
@@ -19,7 +19,17 @@ Classifier: Topic :: Text Processing :: Linguistic
 Requires-Python: >=3.9
 Description-Content-Type: text/markdown
 License-File: LICENSE
+Requires-Dist: transformers
 Dynamic: license-file
 
-# Defuser
-Model defuser helper for HF Transformers 
+
+<div align=center>
+<img width="50%" alt="image" src="https://github.com/user-attachments/assets/f801617b-8959-474a-a565-6b8897e2fcbf" />
+<h1 align="center">Defuser</h1>
+</div>
+
+
+Model defuser helper for HF Transformers >= 5.0. In HF Transformers 5.x releases, many MoE modules became auto-stacked or auto-fused by new modeling code which has benefits but also downsides. 
+
+* Goal is to provide naive module/layer forwarding code for all models supported by HF transformers where run-time weight and structure level optimizations such weight merging, stacking, fusing are reversed so the model is operating in a simple naive state. 
+* There are cases, quantization libraries, where we need to run inference where module input/output needs to be individually captured and this pkg can help complete this task.  

{defuser-0.0.1 → defuser-0.0.2}/Defuser.egg-info/SOURCES.txt

@@ -5,5 +5,9 @@ setup.py
 Defuser.egg-info/PKG-INFO
 Defuser.egg-info/SOURCES.txt
 Defuser.egg-info/dependency_links.txt
+Defuser.egg-info/requires.txt
 Defuser.egg-info/top_level.txt
-defuser/__init__.py
+defuser/__init__.py
+defuser/defuser.py
+defuser/logger.py
+tests/test_convert_model.py

defuser-0.0.2/Defuser.egg-info/requires.txt

@@ -0,0 +1 @@
+transformers

{defuser-0.0.1 → defuser-0.0.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: Defuser
-Version: 0.0.1
+Version: 0.0.2
 Summary: Model defuser helper for HF Transformers.
 Author-email: ModelCloud <qubitium@modelcloud.ai>
 License-Expression: Apache-2.0
@@ -19,7 +19,17 @@ Classifier: Topic :: Text Processing :: Linguistic
 Requires-Python: >=3.9
 Description-Content-Type: text/markdown
 License-File: LICENSE
+Requires-Dist: transformers
 Dynamic: license-file
 
-# Defuser
-Model defuser helper for HF Transformers 
+
+<div align=center>
+<img width="50%" alt="image" src="https://github.com/user-attachments/assets/f801617b-8959-474a-a565-6b8897e2fcbf" />
+<h1 align="center">Defuser</h1>
+</div>
+
+
+Model defuser helper for HF Transformers >= 5.0. In HF Transformers 5.x releases, many MoE modules became auto-stacked or auto-fused by new modeling code which has benefits but also downsides. 
+
+* Goal is to provide naive module/layer forwarding code for all models supported by HF transformers where run-time weight and structure level optimizations such weight merging, stacking, fusing are reversed so the model is operating in a simple naive state. 
+* There are cases, quantization libraries, where we need to run inference where module input/output needs to be individually captured and this pkg can help complete this task.  

defuser-0.0.2/README.md

@@ -0,0 +1,11 @@
+
+<div align=center>
+<img width="50%" alt="image" src="https://github.com/user-attachments/assets/f801617b-8959-474a-a565-6b8897e2fcbf" />
+<h1 align="center">Defuser</h1>
+</div>
+
+
+Model defuser helper for HF Transformers >= 5.0. In HF Transformers 5.x releases, many MoE modules became auto-stacked or auto-fused by new modeling code which has benefits but also downsides. 
+
+* Goal is to provide naive module/layer forwarding code for all models supported by HF transformers where run-time weight and structure level optimizations such weight merging, stacking, fusing are reversed so the model is operating in a simple naive state. 
+* There are cases, quantization libraries, where we need to run inference where module input/output needs to be individually captured and this pkg can help complete this task.  

defuser-0.0.2/defuser/__init__.py

@@ -0,0 +1,6 @@
+# SPDX-FileCopyrightText: 2026 ModelCloud.ai
+# SPDX-FileCopyrightText: 2026 qubitium@modelcloud.ai
+# SPDX-License-Identifier: Apache-2.0
+# Contact: qubitium@modelcloud.ai, x.com/qubitium
+
+from .defuser import convert_hf_model

defuser-0.0.2/defuser/defuser.py

@@ -0,0 +1,14 @@
+# SPDX-FileCopyrightText: 2026 ModelCloud.ai
+# SPDX-FileCopyrightText: 2026 qubitium@modelcloud.ai
+# SPDX-License-Identifier: Apache-2.0
+# Contact: qubitium@modelcloud.ai, x.com/qubitium
+
+from torch import nn
+
+from defuser.utils.hf import apply_modeling_patch
+
+
+def convert_hf_model(model: nn.Module) -> bool:
+    return apply_modeling_patch(model)
+
+__all__ = ["convert_hf_model"]

defuser-0.0.2/defuser/logger.py

@@ -0,0 +1,8 @@
+# SPDX-FileCopyrightText: 2026 ModelCloud.ai
+# SPDX-FileCopyrightText: 2026 qubitium@modelcloud.ai
+# SPDX-License-Identifier: Apache-2.0
+# Contact: qubitium@modelcloud.ai, x.com/qubitium
+
+import logging
+
+logger = logging.getLogger("defuser")

{defuser-0.0.1 → defuser-0.0.2}/pyproject.toml

@@ -9,7 +9,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "Defuser"
-version = "0.0.1"
+version = "0.0.2"
 description = "Model defuser helper for HF Transformers."
 readme = "README.md"
 requires-python = ">=3.9"
@@ -30,6 +30,9 @@ classifiers = [
   "Topic :: Text Processing :: Linguistic"
 ]
 keywords = ["Defuser", "Model", "HF", "Transformers"]
+dependencies = [
+  "transformers",
+]
 
 [project.urls]
 Homepage = "https://github.com/ModelCloud/Defuser"

defuser-0.0.2/tests/test_convert_model.py

@@ -0,0 +1,18 @@
+# SPDX-FileCopyrightText: 2026 ModelCloud.ai
+# SPDX-FileCopyrightText: 2026 qubitium@modelcloud.ai
+# SPDX-License-Identifier: Apache-2.0
+# Contact: qubitium@modelcloud.ai, x.com/qubitium
+from transformers import AutoModelForCausalLM
+from defuser import convert_hf_model
+
+
+def test_qwen3_moe():
+    from defuser.modeling.unfused_moe.qwen3_moe import LinearQwen3MoeSparseMoeBlock
+
+    model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-30B-A3B")
+
+    assert model.config.model_type == "qwen3_moe"
+
+    converted = convert_hf_model(model)
+    assert converted
+    assert isinstance(model.model.layers[0].mlp, LinearQwen3MoeSparseMoeBlock)

defuser-0.0.1/README.md

@@ -1,2 +0,0 @@
-# Defuser
-Model defuser helper for HF Transformers 

defuser-0.0.1/defuser/__init__.py

@@ -1 +0,0 @@
-