Pinned issues
Issues
- 4
[RFC] Long Term QAT Flow
#987 opened by andrewor14 - 15
is this only for linux?
#957 opened by FurkanGozukara - 0
- 0
Does torch.export preserve the quantize_per_tensor/dequantize_per_tensor ops?
#986 opened by justinchuby - 2
[feat] int8 flash attention
#952 opened by felipemello1 - 0
- 0
Add `aarch64` support for torchao binaries
#922 opened by jerryzh168 - 3
optimizer CPU offload doesn't work outside of CUDA
#958 opened by bghira - 4
[ROCm] Unable to Run FPX Weights
#967 opened by Beinsezii - 3
CUDA environment errors after installing ao
#974 opened by philipbutler - 6
[Feature request] Support bias in linear_nf4
#979 opened by ebsmothers - 1
- 2
- 1
[ao/float8] Discussion of the current Float8 recipe
#972 opened by yanbing-j - 1
RuntimeError: CUDA error: named symbol not found CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1 Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.
#968 opened by kolyan288 - 2
- 10
[MPS] torchao low-bit-precision optim does not expose 'backend' argument to torch.compile
#955 opened by bghira - 3
int8 mixed precision training flaky tests?
#945 opened by jerryzh168 - 10
torchao.ops import error on aarch64 linux
#929 opened by ng-05 - 5
- 10
float8 training with rowwise scaling
#889 opened by vkuzo - 4
[Question] Difference in MXLinear vs MXInferenceLinear grouping direction
#932 opened by Abhijit-2592 - 9
run_vit_b_quant.py runs slower than run_bit_b.py
#898 opened by jerryzh168 - 9
[torchao]NotImplementedError: AffineQuantizedTensor dispatch: attempting to run unimplemented operator/function: aten.permute.default
#890 opened by yanbing-j - 3
- 1
Unify scaled INT8 matmul
#862 opened by gau-nernst - 1
- 1
ValueError: ('Unsupported kind: ', 'FRAGMENT')
#900 opened by loretoparisi - 3
- 1
- 0
[CI] Add CI test for PyTorch 2.5.0rc
#888 opened by gau-nernst - 1
AffineQuantizedTensor dispatch: attempting to run unimplemented operator/function: `aten.linear.default`
#875 opened by agunapal - 6
Plans for 8da4w quantization
#883 opened by sanchitintel - 2
Out Of Memory Error
#879 opened by bianxuxuxu - 5
fpx test failing on main
#792 opened by msaroufim - 1
Update Eval scripts
#853 opened by drisspg - 1
Expected Tensor argument scales to have dtype torch.bfloat16, but got torch.float32 instead
#876 opened by agunapal - 1
AO dtype composability tracker
#844 opened by msaroufim - 3
Device error on 8/31 nightlies
#795 opened by ebsmothers - 0
Create report for Autoquant
#855 opened by drisspg - 2
quantization subclass typo
#846 opened by YihengBrianWu - 4
- 0
Add MyPy support
#856 opened by drisspg - 1
Add Autoquant Cache
#854 opened by drisspg - 0
- 0
- 2
installing torchao from source does not work with cuda on latest torchao + latest PT nightly
#810 opened by vkuzo - 7
- 1
- 1
placeholder for float8 README.md images
#793 opened by vkuzo