onnx-diagnostic 0.8.9__py3-none-any.whl → 0.8.11__py3-none-any.whl

onnx_diagnostic/torch_export_patches/serialization/transformers_impl.py

@@ -1,6 +1,7 @@
 import itertools
 from typing import Any, Callable, List, Set, Tuple
 import torch
+import transformers.cache_utils
 from transformers.cache_utils import Cache, DynamicCache, EncoderDecoderCache, StaticCache
 
 try:
@@ -22,22 +23,43 @@ from transformers.modeling_outputs import BaseModelOutput
 from ...helpers.cache_helper import make_dynamic_cache, make_static_cache, CacheKeyValue
 from . import make_serialization_function_for_dataclass
 
-
 SUPPORTED_DATACLASSES: Set[type] = set()
 WRONG_REGISTRATIONS = {
     DynamicCache: "4.50",
     BaseModelOutput: None,
 }
+SHORTEN_LAYER_NAMES = {
+    "DynamicLayer": "D",
+    "DynamicSlidingWindowLayer": "W",
+    "StaticLayer": "S",
+    "StaticSlidingWindowLayer": "X",
+    "D": "DynamicLayer",
+    "W": "DynamicSlidingWindowLayer",
+    "S": "StaticLayer",
+    "X": "StaticSlidingWindowLayer",
+}
 
 
 def _flatten_key_value_cache(cache: Cache) -> Tuple[List[Any], torch.utils._pytree.Context]:
     ca = CacheKeyValue(cache)
     flat = list(itertools.chain.from_iterable(zip(ca.key_cache, ca.value_cache)))
-    keys = list(
-        itertools.chain.from_iterable(
-            (f"key_{i}", f"value_{i}") for i in range(len(ca.key_cache))
+    unique = set(ca.cls_layers) if ca.cls_layers else None
+    if (
+        cache.__class__.__name__ != "DynamicCache"
+        or unique is None
+        or (len(unique) == 1 and unique.pop().__name__ == "DynamicLayer")
+    ):
+        keys = list(
+            itertools.chain.from_iterable(
+                (f"key_{i}", f"value_{i}") for i in range(len(ca.key_cache))
+            )
         )
-    )
+        return flat, keys
+
+    keys = []
+    for i in range(len(ca.key_cache)):
+        letter = SHORTEN_LAYER_NAMES[ca.cls_layers[i].__name__]
+        keys.extend([f"key_{letter}{i}", f"value_{letter}{i}"])
     return flat, keys
 
 
@@ -55,7 +77,20 @@ def _unflatten_cache(
     output_type=None,
 ) -> DynamicCache:
     """Restores a :class:`transformers.cache_utils.DynamicCache` from python objects."""
-    res = make_cache(list(zip(values[::2], values[1::2])))
+    expected = list(
+        itertools.chain.from_iterable(
+            (f"key_{i}", f"value_{i}") for i in range(len(values) // 2)
+        )
+    )
+    if expected == context:
+        res = make_cache(list(zip(values[::2], values[1::2])))
+    else:
+        cls_layer_names = [SHORTEN_LAYER_NAMES[name.split("_")[1][0]] for name in context][::2]
+        cls_layers = [
+            getattr(transformers.cache_utils, cls_name) for cls_name in cls_layer_names
+        ]
+        res = make_cache(list(zip(values[::2], values[1::2])), cls_layers=cls_layers)
+
     assert output_type is None or isinstance(
         res, output_type
     ), f"Type mismatch between {output_type} (expected) and {type(res)}"
@@ -71,14 +106,6 @@ def flatten_dynamic_cache(
     dynamic_cache: DynamicCache,
 ) -> Tuple[List[Any], torch.utils._pytree.Context]:
     """Serializes a :class:`transformers.cache_utils.DynamicCache` with python objects."""
-    assert (
-        not hasattr(dynamic_cache, "layers")
-        or not dynamic_cache.layers
-        or all(lay.__class__.__name__ == "DynamicLayer" for lay in dynamic_cache.layers)
-    ), (
-        f"The serialization does not work yet on other layers "
-        f"than DynamicLayer, but layers={[lay.__class__ for lay in dynamic_cache.layers]}"
-    )
     return _flatten_key_value_cache(dynamic_cache)
 
 
@@ -86,14 +113,6 @@ def flatten_with_keys_dynamic_cache(
     dynamic_cache: DynamicCache,
 ) -> Tuple[List[Tuple[torch.utils._pytree.KeyEntry, Any]], torch.utils._pytree.Context]:
     """Serializes a :class:`transformers.cache_utils.DynamicCache` with python objects."""
-    assert (
-        not hasattr(dynamic_cache, "layers")
-        or not dynamic_cache.layers
-        or all(lay.__class__.__name__ == "DynamicLayer" for lay in dynamic_cache.layers)
-    ), (
-        f"The serialization does not work yet on other layers "
-        f"than DynamicLayer, but layers={[lay.__class__ for lay in dynamic_cache.layers]}"
-    )
     return _flatten_with_keys_cache(dynamic_cache)
 
 
@@ -161,7 +180,9 @@ def unflatten_static_cache(
 ) -> StaticCache:
     """Restores a :class:`transformers.cache_utils.StaticCache` from python objects."""
     return _unflatten_cache(
-        lambda *args: make_static_cache(*args, max_cache_len=values[0].shape[2]),
+        lambda *args, **kwargs: make_static_cache(
+            *args, max_cache_len=values[0].shape[2], **kwargs
+        ),
         values,
         context,
         output_type=output_type,

onnx_diagnostic/torch_models/code_sample.py

@@ -8,11 +8,9 @@ from .hghub.model_inputs import _preprocess_model_id
 from .hghub import get_untrained_model_with_inputs
 from .validate import filter_inputs, make_patch_kwargs
 
-
 CODE_SAMPLES = {
     "imports": "from typing import Any\nimport torch",
-    "get_model_with_inputs": textwrap.dedent(
-        """
+    "get_model_with_inputs": textwrap.dedent("""
     def get_model_with_inputs(
         model_id:str,
         subfolder: str | None = None,
@@ -57,8 +55,7 @@ CODE_SAMPLES = {
             if device:
                 data["model"] = data["model"].to(device)
         return data["model"]
-    """
-    ),
+    """),
 }
 
 
@@ -198,7 +195,7 @@ def code_sample(
         this is not always possible
     :param use_pretrained: use the trained model, not the untrained one
     :param optimization: optimization to apply to the exported model,
-        depend on the the exporter
+        depend on the exporter
     :param quiet: if quiet, catches exception if any issue
     :param patch: applies patches (``patch_transformers=True, path_diffusers=True``)
         if True before exporting
@@ -326,11 +323,9 @@ def code_sample(
         imports,
         cache_import,
         CODE_SAMPLES["get_model_with_inputs"],
-        textwrap.dedent(
-            f"""
+        textwrap.dedent(f"""
             model = get_model_with_inputs({model_args})
-                        """
-        ),
+                        """),
         f"inputs = {input_code}",
         exporter_code,
     ]

onnx_diagnostic/torch_models/hghub/hub_data.py

@@ -10,8 +10,7 @@ __data_arch_values__ = {
     "ResNetForImageClassification": dict(image_size=224),
 }
 
-__data_arch__ = textwrap.dedent(
-    """
+__data_arch__ = textwrap.dedent("""
     architecture,task
     ASTModel,feature-extraction
     AutoencoderKL,image-to-image
@@ -166,8 +165,7 @@ __data_arch__ = textwrap.dedent(
     YolosModel,image-feature-extraction
     Alibaba-NLP/gte-large-en-v1.5,sentence-similarity
     emilyalsentzer/Bio_ClinicalBERT,fill-mask
-    nvidia/Cosmos-Predict2-2B-Video2World//transformer,image-to-video"""
-)
+    nvidia/Cosmos-Predict2-2B-Video2World//transformer,image-to-video""")
 
 __data_tasks__ = [
     "audio-classification",

onnx_diagnostic/torch_models/hghub/hub_data_cached_configs.py

@@ -143,6 +143,7 @@ def _ccached_microsoft_phi2():
             "transformers_version": "4.51.0.dev0",
             "use_cache": true,
             "vocab_size": 51200,
+            "pad_token_id": 0,
         }
     )
 
@@ -193,8 +194,7 @@ def _ccached_hf_internal_testing_tiny_random_beitforimageclassification():
 
 def _ccached_hf_internal_testing_tiny_random_convnext():
     "hf-internal-testing/tiny-random-convnext"
-    t64 = textwrap.dedent(
-        """
+    t64 = textwrap.dedent("""
     ewogICJhcmNoaXRlY3R1cmVzIjogWwogICAgIkNvbnZOZXh0Rm9ySW1hZ2VDbGFzc2lmaWNhdGlvbiIKI
     CBdLAogICJkZXB0aHMiOiBbCiAgICAzLAogICAgMywKICAgIDksCiAgICAzCiAgXSwKICAiZHJvcF9wYX
     RoX3JhdGUiOiAwLjAsCiAgImhpZGRlbl9hY3QiOiAiZ2VsdSIsCiAgImhpZGRlbl9zaXplcyI6IFsKICA
@@ -1344,8 +1344,7 @@ def _ccached_hf_internal_testing_tiny_random_convnext():
     F0Y2hfc2l6ZSI6IDQsCiAgInN0YWdlX25hbWVzIjogWwogICAgInN0ZW0iLAogICAgInN0YWdlMSIsCiA
     gICAic3RhZ2UyIiwKICAgICJzdGFnZTMiLAogICAgInN0YWdlNCIKICBdLAogICJ0b3JjaF9kdHlwZSI6
     ICJmbG9hdDMyIiwKICAidHJhbnNmb3JtZXJzX3ZlcnNpb24iOiAiNC41MS4wLmRldjAiCn0K
-    """.strip()
-    )
+    """.strip())
     js = base64.b64decode(t64.encode("utf-8"))
     kwargs = json.loads(js)
     return transformers.ConvNextConfig(**kwargs)
@@ -1816,8 +1815,7 @@ def _ccached_sshleifer_tiny_marian_en_de():
 
 def _ccached_hf_internal_testing_tiny_random_maskformerforinstancesegmentation():
     "hf-internal-testing/tiny-random-MaskFormerForInstanceSegmentation"
-    t64 = textwrap.dedent(
-        """
+    t64 = textwrap.dedent("""
     ewogICJhcmNoaXRlY3R1cmVzIjogWwogICAgIk1hc2tGb3JtZXJGb3JJbnN0YW5jZVNlZ21lbnRhdGlvb
     iIKICBdLAogICJiYWNrYm9uZSI6IG51bGwsCiAgImJhY2tib25lX2NvbmZpZyI6IHsKICAgICJhdHRlbn
     Rpb25fcHJvYnNfZHJvcG91dF9wcm9iIjogMC4wLAogICAgImRlcHRocyI6IFsKICAgICAgMSwKICAgICA
@@ -1865,8 +1863,7 @@ def _ccached_hf_internal_testing_tiny_random_maskformerforinstancesegmentation()
     b2F0MzIiLAogICJ0cmFuc2Zvcm1lcnNfdmVyc2lvbiI6ICI0LjUxLjAuZGV2MCIsCiAgInVzZV9hdXhpb
     GlhcnlfbG9zcyI6IGZhbHNlLAogICJ1c2VfcHJldHJhaW5lZF9iYWNrYm9uZSI6IGZhbHNlLAogICJ1c2
     VfdGltbV9iYWNrYm9uZSI6IGZhbHNlCn0K
-    """.strip()
-    )
+    """.strip())
     js = base64.b64decode(t64.encode("utf-8"))
     kwargs = json.loads(js)
     return transformers.MaskFormerConfig(**kwargs)
@@ -1909,8 +1906,7 @@ def _ccached_echarlaix_tiny_random_mistral():
 
 def _ccached_hf_internal_testing_tiny_random_mobilevit():
     "hf-internal-testing/tiny-random-mobilevit"
-    t64 = textwrap.dedent(
-        """
+    t64 = textwrap.dedent("""
     ewogICJhcmNoaXRlY3R1cmVzIjogWwogICAgIk1vYmlsZVZpVEZvckltYWdlQ2xhc3NpZmljYXRpb24iC
     iAgXSwKICAiYXNwcF9kcm9wb3V0X3Byb2IiOiAwLjEsCiAgImFzcHBfb3V0X2NoYW5uZWxzIjogMjU2LA
     ogICJhdHJvdXNfcmF0ZXMiOiBbCiAgICA2LAogICAgMTIsCiAgICAxOAogIF0sCiAgImF0dGVudGlvbl9
@@ -3063,8 +3059,7 @@ def _ccached_hf_internal_testing_tiny_random_mobilevit():
     iAgIm91dHB1dF9zdHJpZGUiOiAzMiwKICAicGF0Y2hfc2l6ZSI6IDIsCiAgInFrdl9iaWFzIjogdHJ1ZS
     wKICAic2VtYW50aWNfbG9zc19pZ25vcmVfaW5kZXgiOiAyNTUsCiAgInRvcmNoX2R0eXBlIjogImZsb2F
     0MzIiLAogICJ0cmFuc2Zvcm1lcnNfdmVyc2lvbiI6ICI0LjUxLjAuZGV2MCIKfQo=
-    """.strip()
-    )
+    """.strip())
     js = base64.b64decode(t64.encode("utf-8"))
     kwargs = json.loads(js)
     return transformers.MobileViTConfig(**kwargs)

onnx_diagnostic/torch_models/untrained/llm_phi2.py

@@ -53,6 +53,7 @@ def get_phi2(
         "transformers_version": "4.37.0",
         "use_cache": True,
         "vocab_size": 51200,
+        "pad_token_id": 0,
     }
     config.update(**kwargs)
     conf = transformers.PhiConfig(**config)

onnx_diagnostic/torch_models/validate.py

@@ -871,7 +871,7 @@ def validate_model(
         this is not always possible
     :param use_pretrained: use the trained model, not the untrained one
     :param optimization: optimization to apply to the exported model,
-        depend on the the exporter
+        depend on the exporter
     :param quiet: if quiet, catches exception if any issue
     :param patch: applies patches (``patch_transformers=True, path_diffusers=True``)
         if True before exporting

onnx_diagnostic/torch_onnx/compare.py

@@ -5,7 +5,6 @@ import numpy as np
 import onnx
 from ..helpers.onnx_helper import onnx_dtype_name
 
-
 _NOT_SO_FAR_OPS = [
     {"MatMul", "Gemm", "FusedMatMul"},
     {"Conv", "FusedConv"},

onnx_diagnostic/torch_onnx/runtime_info.py

@@ -2,7 +2,7 @@ import enum
 from typing import Any, Dict, List, Optional, Set, Tuple, Union
 import onnx
 import torch
-from ..api import TensorLike
+from ..typing import TensorLike
 from ..helpers import string_type
 from ..helpers.onnx_helper import get_hidden_inputs
 

onnx_diagnostic/torch_onnx/sbs.py

@@ -374,7 +374,7 @@ def _preparation_with_fx_graph(
                 positions[n] = dict(fx=i)
         if node.op == "placeholder":
             if node.name in placeholders_to_state_dict:
-                # This a weight.
+                # This is a weight.
                 placeholders[node.name] = ep_state_dict[placeholders_to_state_dict[node.name]]
                 torch_results[node.name] = placeholders[node.name]
                 assert isinstance(torch_results[node.name], torch.Tensor), (

onnx_diagnostic/torch_onnx/sbs_dataclasses.py

@@ -120,8 +120,7 @@ class ReplayConfiguration:
             rc = ReplayConfiguration(dump_folder="unused")
             print(rc.get_replay_code())
         """
-        return textwrap.dedent(
-            """
+        return textwrap.dedent("""
             import onnx
             import torch
             from onnx_diagnostic.helpers import max_diff, string_diff, string_type
@@ -225,8 +224,7 @@ class ReplayConfiguration:
                     sess.run(None, ep_feeds)
                 obj = prof.key_averages()
                 print(obj.table())
-            """
-        )
+            """)
 
     def dump(
         self,

onnx_diagnostic/typing.py

@@ -0,0 +1,15 @@
+from typing import Any, Dict, List, Protocol, Tuple, runtime_checkable
+
+
+@runtime_checkable
+class TensorLike(Protocol):
+    @property
+    def shape(self) -> Tuple[int, ...]: ...
+    @property
+    def dtype(self) -> object: ...
+
+
+@runtime_checkable
+class InferenceSessionLike(Protocol):
+    def __init__(self, model: Any, **kwargs): ...
+    def run(self, feeds: Dict[str, TensorLike]) -> List[TensorLike]: ...

{onnx_diagnostic-0.8.9.dist-info → onnx_diagnostic-0.8.11.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: onnx-diagnostic
-Version: 0.8.9
+Version: 0.8.11
 Summary: Tools to help converting pytorch models into ONNX.
 Home-page: https://github.com/sdpython/onnx-diagnostic
 Author: Xavier Dupré

{onnx_diagnostic-0.8.9.dist-info → onnx_diagnostic-0.8.11.dist-info}/RECORD

@@ -1,48 +1,50 @@
-onnx_diagnostic/__init__.py,sha256=9vcutQbxSFR9MMn3ikZxIH3SQ1tgKjRC1yhyumWJ0Go,173
+onnx_diagnostic/__init__.py,sha256=Rvw2q5b1Qw5vuqQ3xWEgU1oyUhLffzhnDfncZoOx6Qc,174
 onnx_diagnostic/__main__.py,sha256=YmyV_Aq_ianDlHyKLHMa6h8YK3ZmFPpLVHLKjM91aCk,79
-onnx_diagnostic/_command_lines_parser.py,sha256=g_udwHBHmY6X_d41Qby_DqMpEHL1p9GfUhJGBCihl8c,57784
-onnx_diagnostic/api.py,sha256=BhCl_yCd78N7TlVtPOHjeYv1QBEy39TjZ647rcHqLh0,345
+onnx_diagnostic/_command_lines_parser.py,sha256=nR-WI-15da9e0o-z5mIKezCH4dGFU66gCbyuAFu2FzY,59315
 onnx_diagnostic/doc.py,sha256=-w2qaes0G0TM0PQFTzBVcJPh0r_IR6Vd2j2fj9iV65Y,10478
-onnx_diagnostic/ext_test_case.py,sha256=A6BkrRm-QbvM8A-qRRMLt9o9ZO6wMXE9jrotggjpGfE,50460
+onnx_diagnostic/ext_test_case.py,sha256=5WQrMdILgr-j8hrqBs7Vi6IFMD3TXNHD7Dlr9K5y1es,51495
+onnx_diagnostic/typing.py,sha256=okLjQGA_ikThoMQ1ikjeu-pCi9FiaO8uhSoKz1ufCQ8,409
 onnx_diagnostic/ci_models/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 onnx_diagnostic/ci_models/ci_helpers.py,sha256=lblOF7z2kLcCRAwMOdqp-Tz1EL1oBywHfVokhqiTQRg,15592
-onnx_diagnostic/ci_models/export_phi4_mm.py,sha256=DH225jXhOG_S8KZE9Kn7FGNE4VqFyc9ZDnG-qxZn-hk,41668
+onnx_diagnostic/ci_models/export_phi4_mm.py,sha256=hekQikmMJi4ihUf494zXVTgMFm0l7RXyfklzroObsYs,41646
 onnx_diagnostic/ci_models/export_qwen25_vl.py,sha256=_rYPr8PPraWizr2MPcGuYjrJ55ilJOyKl8kg0wq4L90,20405
 onnx_diagnostic/export/__init__.py,sha256=yEIoWiOeTwBsDhyYt2fTKuhtA0Ya1J9u9ZzMTOTWaWs,101
-onnx_diagnostic/export/api.py,sha256=iowYszK9ygdciW6P-C0isvW8ZAJW4yvI-3bjiWr2WAE,42945
+onnx_diagnostic/export/api.py,sha256=CL2mwRAJU5tv4ujdlLgMLiC5GoqqeheTW4-xdVRry2M,43172
 onnx_diagnostic/export/cf_simple_loop_for.py,sha256=OHPGQc9AC-0TBtCYpP6cm-iHP9gmNt8WYRrPlO9ewlc,21158
 onnx_diagnostic/export/control_flow_onnx.py,sha256=izGlctqQANrHzSxPMbT7hoauNbnIBdx6hb8ry7HtVmM,18263
 onnx_diagnostic/export/dynamic_shapes.py,sha256=_3xnWXa4n7fTkAT0NSLyLkcEJqm896wHKA8ilEyxvm0,44746
 onnx_diagnostic/export/onnx_plug.py,sha256=U13fL0BjnhMzcDGxaAOqM4TQte5Z4zKDg4ESS0iktjM,22704
 onnx_diagnostic/export/shape_helper.py,sha256=K3y8-vFXYGg5R1tgeVOm_RjCZ8-yyrvYkQ4b3ILM5H4,10990
-onnx_diagnostic/export/validate.py,sha256=_PGUql2DJhIgGKo0WjTGUc5AgsZUx8fEs00MePy-w98,6043
+onnx_diagnostic/export/validate.py,sha256=t8NZjgLqLjabKpUDjO5A5nBk_-BAsN1cdq-6UJ0Jm88,6127
 onnx_diagnostic/helpers/__init__.py,sha256=GJ2GT7cgnlIveVUwMZhuvUwidbTJaKv8CsSIOpZDsJg,83
 onnx_diagnostic/helpers/_log_helper.py,sha256=OTwQH0OIxs9B6nrSvR7MoxMimSw_8mU0mj133NvLk5o,16832
-onnx_diagnostic/helpers/args_helper.py,sha256=SRWnqC7EENg09RZlA50B_PcdiIhdbgA4C3ACfzl5nMs,4419
-onnx_diagnostic/helpers/bench_run.py,sha256=Vvzb7Wy0baIT5O0dx4RKQTx-5V08PiHxPJh6XPkY-lU,16544
-onnx_diagnostic/helpers/cache_helper.py,sha256=aKy566IGuNTZI14Fk3BlByxPYzuU-401eiXY8GN_UWE,31895
+onnx_diagnostic/helpers/args_helper.py,sha256=qHxXJAM4ovBSYeufVVGMRQwrOW2AuP_HFQgqgU-QBQM,4463
+onnx_diagnostic/helpers/bench_run.py,sha256=9T26icFQOSo3HB1gEw31bf-mD5BrVATrRDhuyWESxz8,16543
+onnx_diagnostic/helpers/cache_helper.py,sha256=p7BB7Ba8n2hAPtpvhEuWJ9h9nQF3TlMQZPiMjJZQe70,32029
 onnx_diagnostic/helpers/config_helper.py,sha256=cWRETgFhZ7tayIZPnMqF8BF5AvTU64G2BMqyzgO7lzs,5670
-onnx_diagnostic/helpers/doc_helper.py,sha256=pl5MZd3_FaE8BqQnqoBuSBxoNCFcd2OJd3eITUSku5c,5897
+onnx_diagnostic/helpers/doc_helper.py,sha256=9rigmq36D20yCKG5VkQnODiWMG2eJ0p22UdHew66UrQ,5983
 onnx_diagnostic/helpers/dot_helper.py,sha256=hwgTJsbsUv0qq7euyPDnc1NsBZDGOwv32JXSZxIHJkE,8118
 onnx_diagnostic/helpers/fake_tensor_helper.py,sha256=59046wDIw84or6PJxLaa2CFqaWT7Y3mpYr-BB2shcBE,12027
-onnx_diagnostic/helpers/graph_helper.py,sha256=hevQT5a7_QuriVPQcbT5qe18n99Doyl5h3-qshx1-uk,14093
+onnx_diagnostic/helpers/graph_helper.py,sha256=ailTFdJu1Kir1VQ4GDsm9bUDIevZmgsJVJBGu7CFbCM,14100
 onnx_diagnostic/helpers/helper.py,sha256=7rBu2mQVxQxXrW1YuK433Lp68lvgNOhEqUNkuv9GxJw,66454
-onnx_diagnostic/helpers/log_helper.py,sha256=3mWQd-nLKCctKZt9N8SpoWgLC8O7YdNQ2pfW5QXYWDQ,93232
-onnx_diagnostic/helpers/memory_peak.py,sha256=M3m4_thWFIwP5HytbJYEqaijXIv5v5BW_vlcJowIYI4,6434
-onnx_diagnostic/helpers/mini_onnx_builder.py,sha256=jR2lkRZEQ0N30H0FqeBwaxJd_w_6kyxFagrnulqFjhE,23883
-onnx_diagnostic/helpers/model_builder_helper.py,sha256=qKIq4Naqq03gk6NfqXLQjSDiKL5FFNc1AEyVX0R8GmA,18540
-onnx_diagnostic/helpers/onnx_helper.py,sha256=f5iyzNIXWS90COkERnEb8txafH19CRYWKdKVY_lF05Q,57801
+onnx_diagnostic/helpers/log_helper.py,sha256=njGse2PeD1muXVYvEMl6dsWTDja76wCWipxrEMAkLiw,94504
+onnx_diagnostic/helpers/memory_peak.py,sha256=kigKbD3cq6ZCkE_nhOgNIkr9zHP9zp7ezZhut8GTrcI,6596
+onnx_diagnostic/helpers/mini_onnx_builder.py,sha256=3IAJRf7bphKN0sY714me7H1UzEA1ejJQNYY7cMbGkCo,23878
+onnx_diagnostic/helpers/model_builder_helper.py,sha256=r8LG3xh_-gwp4RODVM9jhB9HNWvccuNypwygWHoU2jA,18548
+onnx_diagnostic/helpers/onnx_helper.py,sha256=1C-ZOsSS0mPHSMT02MD_0kBXpvJTkbsaom45IPiyu_4,64447
 onnx_diagnostic/helpers/optim_helper.py,sha256=0NiYRwV9GLTub4SEny0dqEhLcajRjEhcgkeBDVr9bGQ,4424
-onnx_diagnostic/helpers/ort_session.py,sha256=Y6iJojJcQtPesy9dRaaxRSeRTLXk0Pdh6-_fzdiPDm0,30779
+onnx_diagnostic/helpers/ort_session.py,sha256=NNgJZ756f2zXCTkJidzoTQ5HiQLlkzQzk8hYQp9b0qM,30778
 onnx_diagnostic/helpers/rt_helper.py,sha256=YhVB7v5iR_UT9IqCeqyg58IEIx18f_TnCO6u6GhUciQ,38711
 onnx_diagnostic/helpers/torch_fx_graph_helper.py,sha256=7xFe4svdbr4gV3OTNcx8eJejjDyHAv4hD_RNNKSxL0c,6571
-onnx_diagnostic/helpers/torch_helper.py,sha256=H_rNXq4G78-45pJ2vsO0b_VxYVpSp6DVrOKaal6-VXo,39346
+onnx_diagnostic/helpers/torch_helper.py,sha256=zy2F3pSKnby_0gHH-1rBiE1dvu6NVp_pLNro560ATgs,39449
+onnx_diagnostic/investigate/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+onnx_diagnostic/investigate/input_observer.py,sha256=O42MxwjlrFBgb8nmg1UGLRYUBxoVv0JsSkUARxl0HQk,13882
 onnx_diagnostic/reference/__init__.py,sha256=rLZsxOlnb7-81F2CzepGnZLejaROg4JvgFaGR9FwVQA,208
-onnx_diagnostic/reference/evaluator.py,sha256=RzNzjFDeMe-4X51Tb22N6aagazY5ktNq-mRmPcfY5EU,8848
-onnx_diagnostic/reference/ort_evaluator.py,sha256=q7Dn0yC3LPadlfRnhiRzVn32k9ma_IivdYyhyecgNgc,33930
+onnx_diagnostic/reference/evaluator.py,sha256=5PcPL_5xuEmm4j16KK0o2N5UkdUpuTSxQYc48MPWLpw,8847
+onnx_diagnostic/reference/ort_evaluator.py,sha256=J3keuxFGnQIKP9V80U61y9rpU3zE26RR7HPozeaeiYo,33929
 onnx_diagnostic/reference/quantized_tensor.py,sha256=5u67uS2uGacdMD5VYCbpojNjiesDlV_kO0fAJ0vUWGE,1098
-onnx_diagnostic/reference/report_results_comparison.py,sha256=OsyQN8EHZZoj97u74RQP-7WFpebPOso5GEDpdkLWu6M,3645
-onnx_diagnostic/reference/torch_evaluator.py,sha256=Tx1teWvfGEX5RmkDnI83UiOlo5eBOC72vPhgTWdFUF0,27689
+onnx_diagnostic/reference/report_results_comparison.py,sha256=aldQTYnjfHRdRVOUNldQpCS8U76_KDxwv77-CbNDcNc,3914
+onnx_diagnostic/reference/torch_evaluator.py,sha256=n8BCLcmpi6-ur8JWTHbzuk-dRpJQj42_AwBF94Pdu7U,27921
 onnx_diagnostic/reference/ops/__init__.py,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
 onnx_diagnostic/reference/ops/op_add_add_mul_mul.py,sha256=CXQVtgVrT066gDJFwxL4nDSY4G8r08XNu3EwhWqMapU,1521
 onnx_diagnostic/reference/ops/op_attention.py,sha256=ThALMDF53v3QeG1bohi0bvX2o90HZhGJbbAFOtwEHPE,2027
@@ -73,7 +75,7 @@ onnx_diagnostic/reference/ops/op_slice.py,sha256=yRxfYBs8b7QezyyG9JHCD8MIJHij2qR
 onnx_diagnostic/reference/ops/op_transpose_cast.py,sha256=ifef74rvh0Yvq1Zx51B4mfnISbxV9uRg9DFjkdL1_68,361
 onnx_diagnostic/reference/ops/op_tri_matrix.py,sha256=Yn2gxAyygcwtF5Hjau9ihXDAzul0BAkdqVimVahtFBU,519
 onnx_diagnostic/reference/torch_ops/__init__.py,sha256=fIDMG0KhNYuB4q_57Z5zz0ME6_w8Q0tnqUCooijoOW4,1163
-onnx_diagnostic/reference/torch_ops/_op_run.py,sha256=EEUIwfbRldEFDhULquYhk9x5ZDa9t6f2mKJ1sM__D6A,10517
+onnx_diagnostic/reference/torch_ops/_op_run.py,sha256=qBGaLUWg0wWaTCUlbJq9OpmbHQRiQjtiP7Pbf7LgBXY,10536
 onnx_diagnostic/reference/torch_ops/access_ops.py,sha256=Zfs5OF03PV1CqlCqKI5VV-c4MY3KyQxmO7QZksxQjX8,3274
 onnx_diagnostic/reference/torch_ops/binary_ops.py,sha256=-KxMcCYGDTcZyOss9qU1nU0rmdyg9SdVHJQohseSTcQ,2653
 onnx_diagnostic/reference/torch_ops/controlflow_ops.py,sha256=uOEmzbM4nR2FwZQ8UikwEaHih3yw6T24D_VLYkr5RSU,4518
@@ -81,12 +83,12 @@ onnx_diagnostic/reference/torch_ops/generator_ops.py,sha256=dqqFvhkazVxRUDYhO2t-
 onnx_diagnostic/reference/torch_ops/nn_ops.py,sha256=TeFxQEiTezx9UUzu82ToCLUnUOU56kKv6X81RdZ8UC8,7238
 onnx_diagnostic/reference/torch_ops/other_ops.py,sha256=FnCY60mhdrzrsiHgvN-XpFRHYUpI0gIRqxgVK5J_na0,3995
 onnx_diagnostic/reference/torch_ops/reduce_ops.py,sha256=9gFfraPTQbe_ZEUNCUis1JSmA5dj4tSzjAOpZPJKG4Y,5102
-onnx_diagnostic/reference/torch_ops/sequence_ops.py,sha256=3EiVKpGfN4d1Iry4hgnr3MIJyEEKUrAIDgmRGsUXXa0,2297
+onnx_diagnostic/reference/torch_ops/sequence_ops.py,sha256=POEYPF2sCDM5Ev5w_awpgtcr1XrVhLNva5vgXMUh_UU,2323
 onnx_diagnostic/reference/torch_ops/shape_ops.py,sha256=pJrNR2UB4PlWl6cv4EDl1uGl8YTBUUMQkhJcsh5K4sA,4291
 onnx_diagnostic/reference/torch_ops/unary_ops.py,sha256=dwu6HPr4V_roxu85U3VLTtDLx5bfxKalT_-zlQxZ5wc,1850
 onnx_diagnostic/tasks/__init__.py,sha256=kk-I2tgtb32A_ANh6Ux_u982mA2SrQKO_MDp0KsRi28,2774
 onnx_diagnostic/tasks/automatic_speech_recognition.py,sha256=aMufLDGW005f7aLMZ9alIQtg2s_WIUk5Rd9udS_BZ38,6964
-onnx_diagnostic/tasks/feature_extraction.py,sha256=IS9z9fPNE0hhGUebBfmNZl0twdXobMc7MFKpQB9qZI0,5388
+onnx_diagnostic/tasks/feature_extraction.py,sha256=qfYjK_Pl555eoQjPuk_XnycTKAkr84FtvwrivyRPEc0,5387
 onnx_diagnostic/tasks/fill_mask.py,sha256=5Gt6zlj0p6vuifox7Wmj-TpHXJvPS0CEH8evgdBHDNA,2640
 onnx_diagnostic/tasks/image_classification.py,sha256=nLpBBB1Gkog3Fk6pu2waiHcuQr4ILPptc9FhQ-pn460,4682
 onnx_diagnostic/tasks/image_text_to_text.py,sha256=LFHngpedt4Ws2lj1v0QIqmbTvpV_IbOE05M3o2GRPfs,22289
@@ -103,13 +105,13 @@ onnx_diagnostic/tasks/text_to_image.py,sha256=mOS3Ruosi3hzRMxXLDN7ZkAbi7NnQb7MWw
 onnx_diagnostic/tasks/zero_shot_image_classification.py,sha256=jJCMWuOqGv5ahCfjrcqxuYCJFhTgHV5KUf2yyv2yxYA,4624
 onnx_diagnostic/tasks/data/__init__.py,sha256=uJoemrWgEjI6oA-tMX7r3__x-b3siPmkgqaY7bgIles,401
 onnx_diagnostic/tasks/data/dummies_imagetext2text_generation_gemma3.onnx,sha256=UbtvmWMqcZOKJ-I-HXWI1A6YR6QDaFS5u_yXm5C3ZBw,10299
-onnx_diagnostic/torch_export_patches/__init__.py,sha256=0SaZedwznm1hQUCvXZsGZORV5vby954wEExr5faepGg,720
+onnx_diagnostic/torch_export_patches/__init__.py,sha256=KcBsUXOzw0n-Eo6uFQUUCnJjDsHLnR_J7u-SJu0d6bo,719
 onnx_diagnostic/torch_export_patches/onnx_export_errors.py,sha256=XHYtU7w3vsaTMCuF5X1YtOKxgwL8eEuktXzVZpRz55o,43431
-onnx_diagnostic/torch_export_patches/onnx_export_serialization.py,sha256=zvvtKlqKiC0625vrKP3JNrUHCM3sFV_rPoKWH8Yq9OM,12523
+onnx_diagnostic/torch_export_patches/onnx_export_serialization.py,sha256=KZgr8WbLvmZte1G1hKA0zIlio5ZHz9MKIwGPqNamB6E,12690
 onnx_diagnostic/torch_export_patches/patch_details.py,sha256=UHBo4QTLF3ZgQ4951yYHIQqxOeRYNaG7x56XFcRTtg4,11794
 onnx_diagnostic/torch_export_patches/patch_expressions.py,sha256=VOsv71FsR_UZtxz4-5_VKL2sHQhOkHy9RkPJME2h7UU,3271
 onnx_diagnostic/torch_export_patches/patch_inputs.py,sha256=-TgcyjVzxTb5Y-_ibssTeaA5PFz6FJrV6q84HMUAsJw,8075
-onnx_diagnostic/torch_export_patches/patch_module.py,sha256=1Mn3xdpK1jSdRs6z1C-mJGkfGmD2TNRwLNoPaOW_EFI,40061
+onnx_diagnostic/torch_export_patches/patch_module.py,sha256=Tcgi3osMcTGUsKg03GnbFL7HfZRVncVsJQ50KRHM4Z8,40064
 onnx_diagnostic/torch_export_patches/patch_module_helper.py,sha256=2U0AdyZuU0W54QTdE7tY7imVzMnpQ5091ADNtTCkT8Y,6967
 onnx_diagnostic/torch_export_patches/eval/__init__.py,sha256=15FNHm0Ztz9kuHeXiL0dIT_ABaNtKh_jMRbM8Kt0h7Y,25057
 onnx_diagnostic/torch_export_patches/eval/model_cases.py,sha256=9h4yo9vKiK-E6zaXyAsxXGM-lCjd88ONybA1F3YcTI4,27988
@@ -126,34 +128,34 @@ onnx_diagnostic/torch_export_patches/patches/_patch_transformers_masking_utils.p
 onnx_diagnostic/torch_export_patches/patches/_patch_transformers_qwen2.py,sha256=OxYdlLrwtd_KGHt3E17poduxvWFg-CfGS57-yN1i6gI,3827
 onnx_diagnostic/torch_export_patches/patches/_patch_transformers_qwen2_5.py,sha256=oYz0tr-6KH0DabpgaISytnXAGxQosoA8gV5LpksO4yI,34834
 onnx_diagnostic/torch_export_patches/patches/_patch_transformers_qwen3.py,sha256=cND9Iqo1aKdlX-BXGr9Qlq_Y4EW1L5VWSwZfqYTVazU,4888
-onnx_diagnostic/torch_export_patches/patches/_patch_transformers_rotary_embedding.py,sha256=LAqoL5SWISekZO15G1HIcCkN1JlBxGqb9XbK_eLzalA,16949
+onnx_diagnostic/torch_export_patches/patches/_patch_transformers_rotary_embedding.py,sha256=AQTAAGhpg_Jlzc_26c9rmoZRuhmJcrFX488_W_Z0QD8,16945
 onnx_diagnostic/torch_export_patches/patches/_patch_transformers_sam_mask_decoder.py,sha256=-6TuBm3sLAFEGuW3vRfOTtE5uP6aINFfu7xMnl27Dws,5703
 onnx_diagnostic/torch_export_patches/patches/patch_helper.py,sha256=kK_CGW643iVXxa-m6pttDBS7HTyMQaPypza7iqIInn4,721
-onnx_diagnostic/torch_export_patches/patches/patch_torch.py,sha256=hia-wQRv2LNvUlb-GqML5JGu4a1LEs-lAvooQZefmyM,45196
+onnx_diagnostic/torch_export_patches/patches/patch_torch.py,sha256=IpvyM5yhA3tioyDAwAEe3gZDvbRUwCI_CzKmm76zwyU,45297
 onnx_diagnostic/torch_export_patches/patches/patch_transformers.py,sha256=1W3iKVYx2QT2xJTKlz1UmtjySuwv-rfT5yVL9DjOfzI,3376
 onnx_diagnostic/torch_export_patches/serialization/__init__.py,sha256=BHLdRPtNAtNPAS-bPKEj3-foGSPvwAbZXrHzGGPDLEw,1876
 onnx_diagnostic/torch_export_patches/serialization/diffusers_impl.py,sha256=drq3EH_yjcSuIWYsVeUWm8Cx6YCZFU6bP_1PLtPfY5I,945
-onnx_diagnostic/torch_export_patches/serialization/transformers_impl.py,sha256=kcjMbGfNqftt5X1cP-d153qvK_d5FRRuDyCuieHGieU,10941
+onnx_diagnostic/torch_export_patches/serialization/transformers_impl.py,sha256=divFi7xZcSIm7pERsMg0Eb9a7gDSmXPsdebA_yjoZk4,11530
 onnx_diagnostic/torch_models/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-onnx_diagnostic/torch_models/code_sample.py,sha256=rCDZY64pkn6uIbJJSBuC5TlU_-uleI73I9GlbXtJd54,12923
+onnx_diagnostic/torch_models/code_sample.py,sha256=D0Hv5PJQj2J8Nivoha-RzCdvVhsWYFVFEoG2D2yKm5k,12882
 onnx_diagnostic/torch_models/llms.py,sha256=soyg4yC87ptGoeulJhKqw5opGmuLvH1pn_ZDXZ4Jr8E,90
-onnx_diagnostic/torch_models/validate.py,sha256=JsEMv7aeg9tGGriKZ_CJeqGDfNUmZNdGtPu5jQN9TWY,96739
+onnx_diagnostic/torch_models/validate.py,sha256=THCsMInLdWQ5a5LJbBtziDWHhKMlBXiHORoUr1NVflw,96735
 onnx_diagnostic/torch_models/hghub/__init__.py,sha256=vi1Q7YHdddj1soiBN42MSvJdFqe2_KUoWafHISjwOu8,58
 onnx_diagnostic/torch_models/hghub/hub_api.py,sha256=V3azxUqb7mkmHQ8m5DCgg1WUU2NYBK12USEUy_sfYIA,14709
-onnx_diagnostic/torch_models/hghub/hub_data.py,sha256=6jR8A83cGP4Xw1Wg-q1zzKFpqzoVrybqm0Fm3yurkrE,9030
-onnx_diagnostic/torch_models/hghub/hub_data_cached_configs.py,sha256=Dxa13rsnTQ8eH_BcQvbY2bp1AYFtzuFrJ-J_urrSmeQ,292694
+onnx_diagnostic/torch_models/hghub/hub_data.py,sha256=Xr7bRZk_zCa7sylgZKSRpMcbdFxS4hWxNalM7YKcoDU,9024
+onnx_diagnostic/torch_models/hghub/hub_data_cached_configs.py,sha256=M5B-wa0qi01iivDSgk74QCGE7iy7PvTpgJz7qugj3NM,292683
 onnx_diagnostic/torch_models/hghub/model_inputs.py,sha256=XahJ-m6ajdXg6vFGUOfV5IvFwn-yjAsIOU37nISbBoo,17646
 onnx_diagnostic/torch_models/hghub/model_specific.py,sha256=j50Nu7wddJMoqmD4QzMbNdFDUUgUmSBKRzPDH55TlUQ,2498
 onnx_diagnostic/torch_models/untrained/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-onnx_diagnostic/torch_models/untrained/llm_phi2.py,sha256=y_akbdApi136qHcEQgykwIAYVw0Yfi0lbjb3DNuafaU,3948
+onnx_diagnostic/torch_models/untrained/llm_phi2.py,sha256=HKj0krVNDI--mWHaTzagZ2RWAwe68P2n4JHKuW2ybFI,3975
 onnx_diagnostic/torch_models/untrained/llm_tiny_llm.py,sha256=QXw_Bs2SzfeiQMf-tmtVl83SmVOL4-Um7Qy-f0E48QI,2507
 onnx_diagnostic/torch_onnx/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-onnx_diagnostic/torch_onnx/compare.py,sha256=O0lws4kzn8WAXr8-x-YMPr7oyBC9DtSIs4OfOr4S5-E,12305
-onnx_diagnostic/torch_onnx/runtime_info.py,sha256=u1bD6VXqzBCRmqmbzQtDswaPs1PH_ygr1r-CrcfXpNU,8562
-onnx_diagnostic/torch_onnx/sbs.py,sha256=8okBEIupMgw7TtKc80YFimMtwnY3GchdY05FsA9ooa0,40749
-onnx_diagnostic/torch_onnx/sbs_dataclasses.py,sha256=UctdBjzoPTQG1LS0tZ8A6E9hpoq5HWUYaJLPOPJc9FI,20299
-onnx_diagnostic-0.8.9.dist-info/licenses/LICENSE.txt,sha256=Vv6TXglX6Rc0d-f8aREhayhT-6PMQXEyOmI2NKlUCMc,1045
-onnx_diagnostic-0.8.9.dist-info/METADATA,sha256=cVfclVmpVRwo-9HRCRdVH-YIvWjMWaJloPzmCyN_WOg,6903
-onnx_diagnostic-0.8.9.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-onnx_diagnostic-0.8.9.dist-info/top_level.txt,sha256=KwNkXewmcobM3ZT1DJLVWH6ebJzA5qKg7cWqKfpGNT4,16
-onnx_diagnostic-0.8.9.dist-info/RECORD,,
+onnx_diagnostic/torch_onnx/compare.py,sha256=pXPZ168pT7J_-UOOvsbmS_dCJcwaPjA7GxiLR7Ug8GQ,12304
+onnx_diagnostic/torch_onnx/runtime_info.py,sha256=Mv6WQXNtfNF31zgLm5Hk3x2fTDS6p8UjPh51_Jn4tl0,8565
+onnx_diagnostic/torch_onnx/sbs.py,sha256=xeq4jrMnTeLvZ3UjFlMNpm-VBbP_16fPwTi1_Ujvilo,40752
+onnx_diagnostic/torch_onnx/sbs_dataclasses.py,sha256=WPu9hdz8dIXqvFYy-xtPX9liD_DUVkXOWO_vElOGksY,20277
+onnx_diagnostic-0.8.11.dist-info/licenses/LICENSE.txt,sha256=Vv6TXglX6Rc0d-f8aREhayhT-6PMQXEyOmI2NKlUCMc,1045
+onnx_diagnostic-0.8.11.dist-info/METADATA,sha256=aQ94UbORi24koH-hkkyOjrpCFqnVLYLWXSMrqC_IbOc,6904
+onnx_diagnostic-0.8.11.dist-info/WHEEL,sha256=qELbo2s1Yzl39ZmrAibXA2jjPLUYfnVhUNTlyF1rq0Y,92
+onnx_diagnostic-0.8.11.dist-info/top_level.txt,sha256=KwNkXewmcobM3ZT1DJLVWH6ebJzA5qKg7cWqKfpGNT4,16
+onnx_diagnostic-0.8.11.dist-info/RECORD,,

{onnx_diagnostic-0.8.9.dist-info → onnx_diagnostic-0.8.11.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (80.9.0)
+Generator: setuptools (80.10.1)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

onnx_diagnostic/api.py

@@ -1,15 +0,0 @@
-from typing import Any
-
-
-class TensorLike:
-    """Mocks a tensor."""
-
-    @property
-    def dtype(self) -> Any:
-        "Must be overwritten."
-        raise NotImplementedError("dtype must be overwritten.")
-
-    @property
-    def shape(self) -> Any:
-        "Must be overwritten."
-        raise NotImplementedError("shape must be overwritten.")