mirror of
https://github.com/zebrajr/pytorch.git
synced 2025-12-07 12:21:27 +01:00
2547be5135
On GPU, conv errors if not all its inputs have the same dtype. In the case of autocasting during freezing, what we see is: 1) inputs to conv are casted to half 2) inputs to batchnorm are not casted, so many are still floats 3) we try to fold conv + batchnorm, by finding different weight and bias such that conv(input, new_weight, new_bias) is equivalent to the original conv -> batchnorm. If conv previously had an optional bias, then during freezing we will temporarily create a zero-valued bias as a placeholder for conv_bias. We want to construct it to have the same dtype as the weight input to conv, to avoid errors on GPU. Pull Request resolved: https://github.com/pytorch/pytorch/pull/77295 Approved by: https://github.com/eellison |
||
|---|---|---|
| .. | ||
| init.cpp | ||
| init.h | ||
| module_python.h | ||
| pybind_utils.cpp | ||
| pybind_utils.h | ||
| pybind.h | ||
| python_arg_flatten.cpp | ||
| python_arg_flatten.h | ||
| python_custom_class.cpp | ||
| python_custom_class.h | ||
| python_dict.cpp | ||
| python_dict.h | ||
| python_interpreter.cpp | ||
| python_ir.cpp | ||
| python_ir.h | ||
| python_ivalue.h | ||
| python_list.cpp | ||
| python_list.h | ||
| python_sugared_value.cpp | ||
| python_sugared_value.h | ||
| python_tracer.cpp | ||
| python_tracer.h | ||
| python_tree_views.cpp | ||
| python_tree_views.h | ||
| script_init.cpp | ||
| script_init.h | ||
| update_graph_executor_opt.cpp | ||
| update_graph_executor_opt.h | ||