documentation

Author: xadupre

_doc/conf.py

@@ -115,6 +115,7 @@
     ("py:class", "torch.utils._pytree.Context"),
     ("py:class", "torch.utils._pytree.KeyEntry"),
     ("py:class", "torch.utils._pytree.TreeSpec"),
+    ("py:class", "transformers.GenerationMixin.generate"),
     ("py:class", "transformers.LlamaConfig"),
     ("py:class", "transformers.cache_utils.Cache"),
     ("py:class", "transformers.cache_utils.DynamicCache"),

_doc/examples/plot_export_with_dynamic_cache.py

@@ -5,7 +5,7 @@
 Export with DynamicCache and dynamic shapes
 ===========================================
 
-Every LLMs implemented in :epkg:`trasnformers` use cache.
+Every LLMs implemented in :epkg:`transformers` use cache.
 One of the most used is :class:`transformers.cache_utils.DynamicCache`.
 The cache size is dynamic to cope with the growing context.
 The example shows a tool which determines the dynamic shapes

onnx_diagnostic/export/dynamic_shapes.py

@@ -45,15 +45,15 @@ def forward(self, x, y):
         ds = mi.guess_dynamic_shapes()
         pprint.pprint(ds)
 
-        import pprint
-        import torch
-        from onnx_diagnostic.export import ModelInputs
-
     **kwargs**
 
     .. runpython::
         :showcode:
 
+        import pprint
+        import torch
+        from onnx_diagnostic.export import ModelInputs
+
         class Model(torch.nn.Module):
             def forward(self, x, y):
                 return x + y
@@ -69,15 +69,15 @@ def forward(self, x, y):
         ds = mi.guess_dynamic_shapes()
         pprint.pprint(ds)
 
-        import pprint
-        import torch
-        from onnx_diagnostic.export import ModelInputs
-
     **and and kwargs**
 
     .. runpython::
         :showcode:
 
+        import pprint
+        import torch
+        from onnx_diagnostic.export import ModelInputs
+
         class Model(torch.nn.Module):
             def forward(self, x, y):
                 return x + y