Mu-Chu Lee
Mu-Chu Lee
Error: test/quantization/__test_quantization__/test_quantization#link-tree/torch/ao/quantization/fx/_decomposed.py", line 327, in choose_qparams_tensor_meta assert input.dtype == torch.float32, f"Expecting input to have dtype torch.float32, but got dtype: {input.dtype}"
Summary: We remove the assertion for target_func being cat. The reason is that we have multiple flavors of concat, such as cat/cat.default/cat_slice/cat_slice_cat/... Assertion here is causing multiple times of false...
Summary: Add an additional config to allow buffer mutation. For data that's greater than 2GB, we would need to set it as read-only, otherwise overflow would occur. This is a...