Because of bad network quality, I got error when initializing my model, here is the error message
rasa.core.agent - Could not load model due to Error initializing graph component for node run_LanguageModelFeaturizer1..
This is my config file
- name: LanguageModelFeaturizer
model_name: bert
model_weights: bert-base-chinese
Here is the full debug message:
2022-10-05 10:26:54,640 [DEBUG] Node 'run_LanguageModelFeaturizer1' loading 'LanguageModelFeaturizer.load' and kwargs: '{}'.
2022-10-05 10:26:54,645 [DEBUG] Extracted model to '/tmp/tmpri5j6kw4'.
2022-10-05 10:26:54,650 [DEBUG] Node 'nlu_message_converter' loading 'NLUMessageConverter.load' and kwargs: '{}'.
2022-10-05 10:26:54,650 [DEBUG] Node 'run_custom_tokenizers.jieba_tokenizer_custom.JiebaTokenizerCustom0' loading 'JiebaTokenizerCustom.load' and kwargs: '{}'.
2022-10-05 10:26:54,650 [DEBUG] Node 'run_LanguageModelFeaturizer1' loading 'LanguageModelFeaturizer.load' and kwargs: '{}'.
2022-10-05 10:26:54,830 [DEBUG] Loading Tokenizer and Model for bert
2022-10-05 10:26:54,832 [DEBUG] Loading Tokenizer and Model for bert
2022-10-05 10:26:54,832 [DEBUG] Starting new HTTPS connection (1): huggingface.co:443
2022-10-05 10:26:54,834 [DEBUG] Starting new HTTPS connection (1): huggingface.co:443
2022-10-05 10:26:55,787 [DEBUG] https://huggingface.co:443 "GET /api/models/bert-base-chinese HTTP/1.1" 200 979
2022-10-05 10:26:55,790 [DEBUG] Starting new HTTPS connection (1): huggingface.co:443
2022-10-05 10:26:55,899 [DEBUG] https://huggingface.co:443 "GET /api/models/bert-base-chinese HTTP/1.1" 200 979
2022-10-05 10:26:55,902 [DEBUG] Starting new HTTPS connection (1): huggingface.co:443
2022-10-05 10:26:56,589 [DEBUG] https://huggingface.co:443 "HEAD /bert-base-chinese/resolve/main/vocab.txt HTTP/1.1" 200 0
2022-10-05 10:26:56,593 [DEBUG] Starting new HTTPS connection (1): huggingface.co:443
2022-10-05 10:26:57,448 [DEBUG] https://huggingface.co:443 "HEAD /bert-base-chinese/resolve/main/added_tokens.json HTTP/1.1" 404 0
2022-10-05 10:26:57,451 [DEBUG] Starting new HTTPS connection (1): huggingface.co:443
2022-10-05 10:27:06,157 [DEBUG] Starting new HTTPS connection (1): huggingface.co:443
2022-10-05 10:27:06,948 [DEBUG] https://huggingface.co:443 "HEAD /bert-base-chinese/resolve/main/added_tokens.json HTTP/1.1" 404 0
2022-10-05 10:27:06,951 [DEBUG] Starting new HTTPS connection (1): huggingface.co:443
2022-10-05 10:27:07,690 [ERROR] Could not load model due to Error initializing graph component for node run_LanguageModelFeaturizer1..
I am wondering whether there is a way to use local cache or model storage to avoid loading from huggingface everytime when I run “rasa run”? Thanks for any advice!