-
Environment:
Hardware: Power 10 system (PPC64LE)
OS: Red Hat Enterprise Linux release 9.3 (Plow)
kernel: 5.14.0-362.18.1.el9_3.ppc64le
GH repo: https://github.com/foundation-model-stack/found…
-
### 🚀 Traceable Collectives!
Collective APIs (e.g. all_reduce, all_gather, ...) are used in distributed PyTorch programs, but do not compose cleanly with compilers.
Specifically, torchDynamo a…
-
Sometines the solutions are so simple and almost there in front of you. I was frustrated a lot while working on projects with a local backend that I couldn't change the storage backend to zeo, of m…
-
in stm32 board i can get individual variables iv made in code easily but if i add more than 3 variables (or when mqtt packet has to be published more than 2 times to get all variables ie. in the f…
-
here is the full log when I run `python setup.py install`
Building wheel torch-1.7.0a0+7c50c2f
-- Building version 1.7.0a0+7c50c2f
cmake -GNinja -DBUILD_PYTHON=True -DBUILD_TEST=True -DCMAKE_BUI…
-
Hi! This issue seems to be a feature request or bug report. How to fine tuning with PPO models that doesn't fit one GPU? I'm using FSDP from torch.dist with Accelerate. Unfortunately, ppo_trainer.py a…
-
## Expected Behavior
The `hpx::distributed::barrier::synchronize()` runs successfully.
## Actual Behavior
It gets into an assertion error
```
{file}: /home/jiakuny/workspace/hpx-master/libs…
-
# 🐛 bug report
I'm trying to provide environment variables to select a couple of values on build time, however the env files ([as described here](https://parceljs.org/env.html)) are being ign…
-
**Is your feature request related to a problem? Please describe.**
The debugger has commands for two types of "traces".
- `:bt`, which aligns with the call stack
- `:st`, which shows the scope
…
-
## Vérifs, doc
### `apart_motif_recours`
- [ ] compléter la description `apart_motif_recours` avec APLD
### etat_proc_collective
- [ ] vérifier les différents statuts
- [ ] mieux documenter les…