-
Hi! Greetings from Centor fro High Energy Physics, Tsinghua University.
Currently we succesfully depoyed pytorch with the latest rocm-3.5.0 support on Gentoo. Pytorch uses a python script [build_am…
-
你好,我已经安装jittor,且可以运行example 可是我import jnerf 却出现错误
(/cloud/conda) ➜ work python
Python 3.9.12 (main, Jun 1 2022, 11:38:51)
[GCC 7.5.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credit…
-
使用交叉熵损失函数cross_entropy_loss时,发现没有weight参数,建议加上。
-
## Describe
默认情况下单卡会占用0号GPU,除了设置变量CUDA_VISIBLE_DEVICES外,是否可以提供类似于pytorch的device选项?
-
目前华为MindSpore框架、旷视科技的天元框架竞争压力很大。最近新出了一个算法叫SLIDE,可以利用局部敏感哈希在CPU上运行。
https://github.com/keroro824/HashingDeepLearning
如果能集成这个算法的在jittor上用CPU模式高性能计算,将会是一个巨大的竞争优势。因为省钱不用买卡。
-
Hi,
Nice work! after went thru the code, some questions want to ask,
- How to lower ops like batchnorm to meta ops, tf xla has tf2xla phase to lower big op into a group of meta ops, I don't see …
xqdan updated
4 years ago
-
Centos7环境信息:
gcc 6.1.0
cuda 11.7
python 3.9
[i 0523 18:12:43.006756 44 jit_compiler.cc:28] Load cc_path: /home/aaa/anaconda3/envs/jjitor/bin/g++
terminate called after throwing an instance of '…
-
请问5月30号的研讨会还有回放吗
-
```
(jittor) PS F:\test-code\JittorLLMs> python cli_demo.py pangualpha
WARNING: APEX is not installed, multi_tensor_applier will not be available.
WARNING: APEX is not installed, using torch.nn.Lay…
-