-
### Description
Right now asyncio support in accelerated DAGs uses a slightly different codepath from normal DAGs:
- you need to specify `enable_asyncio=True` in `dag.experimental_compile()`
- you …
-
im getting this error when loadning "**meta-llama/Llama-3.2-90B-Vision-Instruct**", I think this issue only when loading to GPU. I success loaded the same model to CPU without issue.
Anyone knows s…
-
https://github.com/Kuratius/sm64/commit/ace0e206ebab999da123168f7ff2dec482f34351
I tried implementing hardware accelerated fdiv and sqrtf functions, I also tried writing an fmul implementation that…
-
### What behavior of the library made you think about the improvement?
I need to install torch, transformers, accelerate etc. even if I want to use outlines only with llamacpp backend.
Are these d…
-
```
What steps will reproduce the problem?
1. I'm using "rain" effect with sDelay: 45. Leave it sliding during minutes
2. Animation will accelerate, faster and faster, after 3 or 4 minutes I can't
ev…
-
From this record on travis-ci these are our slowest tests:
```
========================== slowest 20 test durations ===========================
41.75s call distributed/tests/test_scheduler.py…
-
Is it possible to use all the VRAM of the GPU (I have 24 gb) to accelerate the generation? Although I check the option to disable CPU offload, it only uses 2 or 3 GB of VRAM.
Thanks in advance.
-
The heterogenous resources within CCE "Learn about GPU-accelerated models" return 404
https://docs.otc.t-systems.com/en-us/productdesc-ecs/ecs_01_0045.html
-
Can the running speed be further optimized? Running efficiency is too slow
-
I would like to have GPU accelerated FFT.
Is it possible?
This will have applications for real time simulations and games.