Martin

icon company CStore icon location Nanjing, JiangSu, China icon location I am a deep learning engineer with a master's degree from Nanjing University of Information Science & Technology.

Results 4 comments of Martin

Could you please share the something v1 dataset?

#### The same problem I had the same problem. Actually, it doesn't matter to me without this dataset, I just want to know how to organize my own dataset.

AutoModelForCausalLM

I also want to know the answer of this question.

如何使用在huggingface下载的离线模型推理glm-10b-chinese?

> 模型应该用`AutoModelForSeq2SeqLM `而不是`AutoModelForCausalLM` > https://github.com/THUDM/GLM#generation 这两个有什么区别吗？

Finetuning Chinese models

我同样比较期待能够支持Encoder-Decoder的Fine-tune，我们有资源可以训练GLM-130B，但是苦于人少和技术能力差距，很难实现，期待你们对Encoder-Decoder架构的支持。 > Hi， > 目前开源的代码还不支持encoder-decoder类型的模型。glm版本我们正在测试代码正确性，这周会放出~