Skip to content

Issues: THUDM/SwissArmyTransformer

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

39 Open 40 Closed
39 Open 40 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

MetaDistributedWebDataset传入的训练数据格式是否有范例参考？
#185 opened Sep 14, 2024 by xiabaoyulo
在使用Deepspeed的zero stage2训练visualglm-6B是，出现最终权重为25G的现象
#182 opened Sep 9, 2024 by corkiyao
ema模型是否正确被保存
#180 opened Aug 10, 2024 by Blueskyvvvvv
1
TypeError: sat.model.transformer.BaseTransformer() got multiple values for keyword argument 'parallel_output'
#179 opened Jul 26, 2024 by deep-practice
33
请问断点续训应该如何设置
#178 opened Apr 11, 2024 by elesun2018
6
transfer_param.py 转换vincuna hf模型成sat模型报错
#177 opened Apr 8, 2024 by Lunatic-Solar
17
NO cogagent？
#175 opened Mar 22, 2024 by Mac0q
2
ModuleNotFoundError: No module named 'localAttention'
#174 opened Mar 20, 2024 by BlueSkyyyyyy
“No backend type associated with device type cpu” when run cli_demo_sat.py
#173 opened Mar 5, 2024 by yileld
5
如果想绕过deepspeed做finetune，可以在train的时候直接model.step()来实现吗？
#172 opened Feb 16, 2024 by cocoshe
2
MixtralMlpMixin()这个函数里面moe只是计算专家的logits但是没看到分发逻辑
#170 opened Feb 9, 2024 by AlenjandroWang
1
AutoModel.from_pretrained()里面不能加载hf版本的权重吗
#169 opened Feb 4, 2024 by AlenjandroWang
1
怎么从断点恢复微调训练
#167 opened Feb 2, 2024 by zoumaguanxin
1
请问针对样本数量不均衡的数据集怎么做样本均衡呢
#163 opened Jan 25, 2024 by lln556
1
deepspeed 分布式训练 loss nan or inf
#161 opened Jan 19, 2024 by JohnTang93
1
Is sat suuport saving checkpoint by using fp16 or bf16?
#160 opened Jan 19, 2024 by xxxwuwq
5
deepspeed分布式训练出现sat ValueError inconsistent
#149 opened Dec 14, 2023 by elesun2018
1
How to embed video encoder module from pytorch？
#148 opened Dec 4, 2023 by zyhzyh88
3
ore.exceptions.ResponseStreamingError
#142 opened Nov 1, 2023 by AnnaYang2020
1
Cannot use torch.compile with SAT
#141 opened Oct 23, 2023 by lijing1996
Fail to load random states from checkpoints saved
#137 opened Sep 25, 2023 by minkowski0125
2
ModuleNotFoundError: No module named 'SwissArmyTransformer'
#133 opened Sep 9, 2023 by B-1368
6
请教一个问题，使用mp_size=2时的loss应该怎么写
#131 opened Aug 24, 2023 by kunden0612
1
模型并行的方式进行lora方式的finetuning要怎么设置呢
#130 opened Aug 22, 2023 by kunden0612
5
During BERT decoding, past_key_values is used to accelerate calculation. Do we have a similar implementation?
#129 opened Aug 10, 2023 by etrigger
1
ProTip! Add no:assignee to see everything that’s not assigned.