Daze Lu

always get acc,acc_norm, perplexity =1 on triviaqa task based on llama2 model

4

I use the following command to run triviaqa task ` lm_eval --model hf \ --model_args pretrained=../llama/models_hf/7B \ --tasks triviaqa \ --num_fewshot 1 \ --device cuda:2 \ --batch_size 8 ` I...

I use Tesla k80 to train the bert_base model by using the command 'GPU1 python train.py train_bert_base', and I change the experiments.conf of bert_base as the following image:  And...

UnicodeEncodeError: 'charmap' codec can't encode characters in position 25-29: character maps to <undefined>

3

不知有没有人遇到过这个问题，在ubuntu上跑的好好的，在一台windows上也跑得好好的，文件open时加的编码是utf-8，在另一台国外电脑上就报这个错 

我使用了一块RTX4090分别训练了中文和英文的OntoNote数据，得到下面的结果 中文：RoBERTa_zh_L12_PyTorch  英文：spanbert_base  使用参数如下： `# Computation limits. max_top_antecedents = 50 max_training_sentences = 11 top_span_ratio = 0.4 max_num_speakers = 20 max_segment_len = 128 # Learning bert_learning_rate = 1e-05 task_learning_rate...

看到不少人问这个问题，在下亲自实践，添加个issue吧 1. 打开这个文件目录，做如下设置  2. 我还执行了`sudo chmod 777 ./scorer.pl`不知是否是必要的，但是两步设置完跑通了  附：windows下没试过，想来更改读写权限，是一样的道理

cpython-310-x86_64-linux-gnu.so: undefined symbol: _ZN3c104cuda9SetDeviceEi

2

Hi, I cloned this project and built the env with README shows, I met tensornvme errors, has anyone met this as well? environment: cuda 12.1 PyTorch 2.5.1 the following error...