CogView
CogView copied to clipboard
THUDM
Text-to-Image generation. The repo for NeurIPS 2021 paper "CogView: Mastering Text-to-Image Generation via Transformers".
Traceback (most recent call last): File "pretrain_gpt2.py", line 829, in Traceback (most recent call last): File "pretrain_gpt2.py", line 829, in main() File "pretrain_gpt2.py", line 742, in main main() File "pretrain_gpt2.py",...
I use model parallel mode and set the parameter model-parallel-size=4, it will save four models. how do I use the four saved models to inference?
Hi, I found your model checkpoint at https://resource.wudaoai.cn/home?ind=2&name=WuDao%20WenHui&id=1399364355975327744 too large for me to run on my PC, as the hidden size of the checkpoint is 2560. Do you have a...
We can read on the readme since March 16th: > News! The [demo](https://agc.platform.baai.ac.cn/CogView/index.html) for a better and faster CogView2 (formal version, March 2022) is available! The lastest model also supports...
Not working by any of these links: https://agc.platform.baai.ac.cn/CogView/index.html https://thudm.github.io/CogView/index.html https://wudao.aminer.cn/CogView/index.html https://lab.aminer.cn/cogview/index.html
使用的环境是由作者提供的docker镜像 使用的显卡是 Tesla P100-PCIE 16GB 在运行./scripts/text2image.sh --debug报错 报错代码如下： `Generate Samples WARNING: No training data specified using world size: 1 and model-parallel size: 1 > using dynamic loss scaling > initializing...
麻烦问两个问题，感谢 1 请问loss在预训练的下降情况是怎么样的，各自初始值是多少呢，最终各自收敛到什么情况呢，finetune大概是什么情况呢 2 lr一般和batch关系比较紧密，那请问finetune过程中你们的batch和lr大概多少呢
不是很明白为什么还需要申请等各种复杂流程，所以到底是否开放呢？清华源也全部失效了
Hi, I just found that the link of pretrained weight Wudao-Wenhui can not be open. Any chance update that link?
Hi, I am very interested in your research. When I reproduced your code, the following problems appeared, and I felt very confused. Traceback (most recent call last): File "generate_samples.py", line...
