3分钟实现推理教程能出个视频么？

FileNotFoundError Traceback (most recent call last)
/code/panguAlpha_pytorch/tools/generate_samples_Pangu.py in
178 if name == "main":
179
--> 180 main()

/code/panguAlpha_pytorch/tools/generate_samples_Pangu.py in main()
127
128 initialize_megatron(extra_args_provider=add_text_generate_args,
--> 129 args_defaults={'tokenizer_type': 'GPT2BPETokenizer'})
130
131 # Set up model and load checkpoint.

/code/panguAlpha_pytorch/megatron/initialize.py in initialize_megatron(extra_args_provider, args_defaults, ignore_unknown_args, allow_no_cuda)
48 set_global_variables(extra_args_provider=extra_args_provider,
49 args_defaults=args_defaults,
---> 50 ignore_unknown_args=ignore_unknown_args)
51
52 # torch.distributed initialization

/code/panguAlpha_pytorch/megatron/global_vars.py in set_global_variables(extra_args_provider, args_defaults, ignore_unknown_args)
68 defaults=args_defaults,
69 ignore_unknown_args=ignore_unknown_args)
---> 70 _ = _build_tokenizer(args)
71 _set_tensorboard_writer(args)
72 _set_adlr_autoresume(args)

/code/panguAlpha_pytorch/megatron/global_vars.py in _build_tokenizer(args)
89 global _GLOBAL_TOKENIZER
90 _ensure_var_is_not_initialized(_GLOBAL_TOKENIZER, 'tokenizer')
---> 91 _GLOBAL_TOKENIZER = build_tokenizer(args)
92 return _GLOBAL_TOKENIZER
93

/code/panguAlpha_pytorch/megatron/tokenizer/tokenizer.py in build_tokenizer(args)
40 # assert args.merge_file is not None
41 # tokenizer = _GPT2BPETokenizer(args.vocab_file, args.merge_file)
---> 42 tokenizer = JIEBATokenizer(args.vocab_file)
43 else:
44 raise NotImplementedError('{} tokenizer is not '

/code/panguAlpha_pytorch/megatron/tokenizer/tokenization_jieba.py in init(self, vocab_file, max_len)
42 model_file = vocab_file + ".model"
43 vocab_file = vocab_file + ".vocab"
---> 44 f = open(vocab_file,'r', encoding='utf-8')
45 lines = f.readlines()
46 self.encoder = {}

FileNotFoundError: [Errno 2] No such file or directory: 'megatron/tokenizer/bpe_4w_pcl/vocab.vocab'

--------------------------------------------------------------------------- FileNotFoundError Traceback (most recent call last) /code/panguAlpha_pytorch/tools/generate_samples_Pangu.py in <module> 178 if __name__ == "__main__": 179 --> 180 main() /code/panguAlpha_pytorch/tools/generate_samples_Pangu.py in main() 127 128 initialize_megatron(extra_args_provider=add_text_generate_args, --> 129 args_defaults={'tokenizer_type': 'GPT2BPETokenizer'}) 130 131 # Set up model and load checkpoint. /code/panguAlpha_pytorch/megatron/initialize.py in initialize_megatron(extra_args_provider, args_defaults, ignore_unknown_args, allow_no_cuda) 48 set_global_variables(extra_args_provider=extra_args_provider, 49 args_defaults=args_defaults, ---> 50 ignore_unknown_args=ignore_unknown_args) 51 52 # torch.distributed initialization /code/panguAlpha_pytorch/megatron/global_vars.py in set_global_variables(extra_args_provider, args_defaults, ignore_unknown_args) 68 defaults=args_defaults, 69 ignore_unknown_args=ignore_unknown_args) ---> 70 _ = _build_tokenizer(args) 71 _set_tensorboard_writer(args) 72 _set_adlr_autoresume(args) /code/panguAlpha_pytorch/megatron/global_vars.py in _build_tokenizer(args) 89 global _GLOBAL_TOKENIZER 90 _ensure_var_is_not_initialized(_GLOBAL_TOKENIZER, 'tokenizer') ---> 91 _GLOBAL_TOKENIZER = build_tokenizer(args) 92 return _GLOBAL_TOKENIZER 93 /code/panguAlpha_pytorch/megatron/tokenizer/tokenizer.py in build_tokenizer(args) 40 # assert args.merge_file is not None 41 # tokenizer = _GPT2BPETokenizer(args.vocab_file, args.merge_file) ---> 42 tokenizer = JIEBATokenizer(args.vocab_file) 43 else: 44 raise NotImplementedError('{} tokenizer is not ' /code/panguAlpha_pytorch/megatron/tokenizer/tokenization_jieba.py in __init__(self, vocab_file, max_len) 42 model_file = vocab_file + ".model" 43 vocab_file = vocab_file + ".vocab" ---> 44 f = open(vocab_file,'r', encoding='utf-8') 45 lines = f.readlines() 46 self.encoder = {} FileNotFoundError: [Errno 2] No such file or directory: 'megatron/tokenizer/bpe_4w_pcl/vocab.vocab'

我照着教程做到最后一步执行运行推理脚本显示报错.
能不能弄个视频完整的演示下，

最后能否请大佬出个3分钟实现训练和生成小说文本的教程视频？
谢谢呀

请问是什么错误？

鹏城.盘古模型的训练、推理、下游任务应用示例等可以进一步参考如下项目，包含详细的使用教程。
https://git.openi.org.cn/PCL-Platform.Intelligence/PanGu-Alpha-GPU
https://git.openi.org.cn/PCL-Platform.Intelligence/PanGu-Alpha-Application。

按着你3分钟的教程一步步执行，显示说没有这个文件，但是文件在那里，这是什么问题？

这个文件显然是存在的，可能是你代码执行路径的问题，你试着把vocab-file改为全路径。

重新验证了一下，教程是没问题。可能是你代码运行路径不对，你仔细按照教程再操作一遍看看？

谢谢已解决

zhaoxiong closed this issue 2 years ago

#20 3分钟实现推理教程能出个视频么？