-
## 🐛 Bug
I am running the asr_wsj recipe. It is training the word_lm (stage 6) since last night but does not produce any output, logging or otherwise.
When I run nvtop or Nvidia-smi the gpus se…
-
## 🐛 Bug
I would like to pre-train the FairSeq Roberta base model using torch-xla on GPU. I am following the FairSeq [Pretraining RoBERTa using your own data tutorial](https://github.com/pytorch/…
-
Aim: improve the maintainability of the project catalogue, allowing it to grow without causing an increasing amount of development overhead. Currently, it is populated via a manually run script that c…
-
#### What is your question?
How do I use trained models in a different machine?
#### What have you tried?
1) In a remote machine, I have **trained a model and generate translations successful…
-
## 🐛 Bug
omegaconf.errors.ConfigAttributeError: Key 'normalize' is not in struct
### To Reproduce
1. Run cmd '....'
fairseq-hydra-train task.data=/mnt/disks2/wav2vec/dataset distributed_tr…
-
hello i have more software files for m600 36V.
One version with 25A and three with 15A.
-
The two java files CMSAuthEnvelopedData and CMSAuthEnvelopedGenerator have only package access and it seems the implementation is not finished yet.
Background is I need to create a 1.2.840.113549.1.9…
-
... Sammlung der Fakten ...
Hardware: Arduino Pro Mini 3.3V 8Mhz
Firmware: V 3.5.0-dev+20210623 SIGNALduino cc1101
Der Fehler ist "reproduzierbar" indem das Unbekannte Signal erneut auftritt.
…
-
Hey,
I got new zigbee wall switches, but they are not supported by deconz.
Is there any chance to add support for those switches?
-
## 🐛 Bug
I have a model that uses Sentencepiece but the new argument parsing system seems to not be able to find the value for sentencepiece-model from the command line.
### To Reproduce
Runn…