shisi-cc

ValueError when running VarLiNGAM

1

When I run VarLiNGAM on Finance dataset (http://www.skleinberg.org/data/FinanceCPT.tar.gz), I meet a ValueError. `df_data = pd.read_csv(datafile) model = VarLiNGAM(lag=3) result = model.create_graph_from_data(df_data)` The error is as follows File "", line 1,...

强化学习使用GradCAM

3

你好，我想尝试在强化学习里用GradCAM，但是遇到一个问题是强化学习里损失函数跟做分类不太一样，我不太确定求GradCAM的梯度时是用强化学习的损失函数，还是像分类一样用最大值做反向，不知道您有没有了解？

没有标签能用GradCAM吗

2

想问一下GradCAM是不是只能用在有标签的情况呀？不然没法反向传播？直觉上好像解释不应该依赖标签才合理？如果没有标签的话应该怎么做可视化呀？

[Feature] <ChatGLM好像不支持pipeline("text-generation")？>

4

### Is your feature request related to a problem? Please describe. _No response_ ### Solutions 想用pipeline做一下模型的推理，但是ChatGLM好像不支持pipeline("text-generation") 除了使用model.chat()，怎么样能让ChatGLM也能够使用pipeline呢？ 报错是 The model 'ChatGLMForConditionalGeneration' is not supported for text-generation. ### Additional context _No...

[BUG] terminate called after throwing an instance of 'std::bad_alloc'

3

**Describe the bug** When I run the code [rlhf with trlx]( https://github.com/CarperAI/trlx/tree/main/examples/summarize_rlhf/sft/train_gptj_summarize.py) using deepspeed with two nodes, I met a strange problem "terminate called after throwing an instance of 'std::bad_alloc'"....

Request for the configuration file

6

I have read the paper and try to implement the models. Could you provide the configuration file (net_config.json) of CNN model in Figure 2 and RNN model in Figure 3?

When I run the preprocess script, I encounter an error. > Traceback (most recent call last): > File "/data/work/TimeLLM/ecg-qa/preprocess_ecgqa.py", line 198, in > main(args) > File "/data/work/TimeLLM/ecg-qa/preprocess_ecgqa.py", line 109, in...

When I run sh scripts/time_series_forecasting/short_term/AutoTimes_M4.sh, I encounter the following error. How can I resolve this? > Traceback (most recent call last): > File "/data/work/TimeLLM/AutoTimes/run.py", line 124, in > exp.train(setting) >...

look forward to publishing pre-trained models

1

I am very interested in this work and look forward to publishing pre-trained models in huggingface repository.