"generate.py" 报错

[yangboz]

环境信息如下：

··· active environment : tf_gpu active env location : /home/server/anaconda3/envs/tf_gpu shell level : 2 user config file : /home/server/.condarc populated config files : /home/server/.condarc conda version : 4.9.2 conda-build version : 3.17.8 python version : 3.7.3.final.0 virtual packages : cuda=11.2=0 glibc=2.27=0 unix=0=0 archspec=1=x86_64 base environment : /home/server/anaconda3 (writable) channel URLs : https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/linux-64 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/noarch https://mirrors.ustc.edu.cn/anaconda/pkgs/free/linux-64 https://mirrors.ustc.edu.cn/anaconda/pkgs/free/noarch https://repo.anaconda.com/pkgs/main/linux-64 https://repo.anaconda.com/pkgs/main/noarch https://repo.anaconda.com/pkgs/r/linux-64 https://repo.anaconda.com/pkgs/r/noarch package cache : /home/server/anaconda3/pkgs /home/server/.conda/pkgs envs directories : /home/server/anaconda3/envs /home/server/.conda/envs platform : linux-64 user-agent : conda/4.9.2 requests/2.21.0 CPython/3.7.3 Linux/5.4.0-65-generic ubuntu/18.04.5 glibc/2.27 UID:GID : 1000:1000 netrc file : None offline mode : False ···

运行信息如下： ··· python generate.py \ --device 0 \ --length=50 --nsamples=4 --prefix=xxx --fast_pattern \ --tokenizer_path prose/vocab.txt \ --model_path prose/ \ --topp 1 \ --temperature 1.0 \ --batch_size 1 ··· 错误信息如下： ··· 2021-02-19 14:46:42.559905: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x564863feabc0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices: 2021-02-19 14:46:42.559923: I tensorflow/compiler/xla/service/service.cc:176] StreamExecutor device (0): GeForce GTX 960M, Compute Capability 5.0 args: Namespace(batch_size=1, device='0', fast_pattern=True, length=50, model_config='config/model_config_small.json', model_path='prose/', no_wordpiece=False, nsamples=4, prefix='xxx', repetition_penalty=1.0, save_samples=False, save_samples_path='.', segment=False, temperature=1.0, tokenizer_path='prose/vocab.txt', topk=8, topp=1.0) 0%| | 0/50 [00:00<?, ?it/s] Traceback (most recent call last): File "generate.py", line 222, in main() File "generate.py", line 182, in main out = generate( File "generate.py", line 117, in generate return fast_sample_sequence(model, context, length, temperature=temperature, top_k=top_k, top_p=top_p, File "generate.py", line 107, in fast_sample_sequence next_token = torch.multinomial(torch.softmax(filtered_logits, dim=-1), num_samples=1) RuntimeError: probability tensor contains either inf, nan or element < 0 ···