-
Hi, do you have an implementation of the Conformer Block in Flashlight you can share?
These are my rough notes for what conformer seems to be: https://gist.github.com/lunixbochs/4d8a8c0ab9be4546933…
-
The tool is [here](https://github.com/Geal/cargo-external-doc). It would be nice to have the extra docs supplied with a project built on docs.rs as well. Not sure of the best way to indicate that extr…
-
Hi,
Just gathering feedback on the title. I think I can implement full MinGW toolchain support for Windows. That means Visual Studio (= 12 GBytes hog) won't required to build CMake.js based modules. …
-
### Bug Description
I'm trying to build flashlight from sources with Cuda backend following steps provided in the documentation but an error occurs in "/flashlight/autograd/backend/cuda/CudnnUtils.cp…
-
### `brew config` output
```shell
HOMEBREW_VERSION: 2.2.2
ORIGIN: https://github.com/Homebrew/brew
HEAD: 13f508d0dc8ada1726ee09a750285d7447ac3df4
Last commit: 1 year, 4 months ago
Core tap ORIGIN…
-
### Question
How to install on Jetson Nano?
on cuda ?
on cpu ?
#### Additional Context
OpenBLAS ok!
cereal ok!
kenlm ok!
arrayfire binary ok! (ArrayFire-v3.5.1_Linux_aarch64.sh not compati…
-
[178/183] : && /usr/bin/c++ -fPIC -g CMakeFiles/fl_asr_decode.dir/flashlight/app/asr/Decode.cpp.o -o bin/asr/fl_asr_decode -L/opt/intel/mkl/lib/intel64 -L/usr/local/cuda/targets/x86_64-linux/lib/st…
-
Pytorch as a functionality to cache the chosen cuDNN kernels when the input dimensions are identical to previous calls:
```
torch.backend.cudnn.benchmark = True
```
This improves throughput of a…
-
RustSec/advisory-db has changed their format and it is failing to parse TOML.
https://github.com/RustSec/advisory-db/commit/64c17acfe3c7e4c57973d54c69d879abf1763154
Note that they have `unaffected…
-
To optimize the obtained error in ML, I have to use smaller batch sizes. For the provide example: neural network, a batch size of 48 is ideal. The overhead increases enormously however.
By recomp…