Emir-Liu
Emir-Liu
The py_maths_tools is still blank 。。waiting
i also meet the same question,but i found the above solution is useless. then,i download the corpus.small as above says https://drive.google.com/file/d/1gdNG92VABX8eWc7JWnU7Y-1wa5cu5-0L/view?usp=sharing. i found all question is solved. i suspect it's...
I used the original parameters .
when using the new version packet,I meet the problem too.And,I change the code in GAIN.py. ```bash # graph_big = dgl.batch_hetero(graphs) graph_big = dgl.batch_hetero(graphs).to('cuda:0') ... # entity_graph_big = dgl.batch(entity_graphs) entity_graph_big =...
i find the parameter only occurs in zh_msra,can you tell me the meaning of the parameter
这个应该是因为，训练的结果是用safetensor保存的，但是预测模型的时候，却使用了bin格式进行加载，导致文件不匹配。 训练方式，codellama模型，不使用deepspeed，使用sft训练 ref:https://github.com/huggingface/safetensors/discussions/397#discussioncomment-10585555
upgrading to gradio 4.0.0 does not work for me.