HumanSignal / label-studio-ml-backend

Configs and boilerplates for Label Studio's Machine Learning backend
Apache License 2.0
587 stars 261 forks source link

Internal Server Error when attempting to train Gliner (docker build) #609

Open maaxlong opened 3 months ago

maaxlong commented 3 months ago

Hello,

I am getting the following error logs when trying to finetune Gliner, based on annotations which have been corrected via Label Studio. I am using docker for the ml backend - it works fine for suggesting annotations, and the model connects successfully, but fails to start training when requested.

2024-08-15 09:04:01 gliner | [2024-08-15 08:04:01,991] [INFO] [model::fit::235] Downloaded 233 labeled tasks from Label Studio 2024-08-15 09:04:02 gliner | [2024-08-15 08:04:01,995] [ERROR] [label_studio_ml.api::log_exception::1414] Exception on /webhook [POST] 2024-08-15 09:04:02 gliner | Traceback (most recent call last): 2024-08-15 09:04:02 gliner | File "/usr/local/lib/python3.11/site-packages/flask/app.py", line 2190, in wsgi_app 2024-08-15 09:04:02 gliner | response = self.full_dispatch_request() 2024-08-15 09:04:02 gliner | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2024-08-15 09:04:02 gliner | File "/usr/local/lib/python3.11/site-packages/flask/app.py", line 1486, in full_dispatch_request 2024-08-15 09:04:02 gliner | rv = self.handle_user_exception(e) 2024-08-15 09:04:02 gliner | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2024-08-15 09:04:02 gliner | File "/usr/local/lib/python3.11/site-packages/flask/app.py", line 1484, in full_dispatch_request 2024-08-15 09:04:02 gliner | rv = self.dispatch_request() 2024-08-15 09:04:02 gliner | ^^^^^^^^^^^^^^^^^^^^^^^ 2024-08-15 09:04:02 gliner | File "/usr/local/lib/python3.11/site-packages/flask/app.py", line 1469, in dispatch_request 2024-08-15 09:04:02 gliner | return self.ensure_sync(self.view_functions[rule.endpoint])(**view_args) 2024-08-15 09:04:02 gliner | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2024-08-15 09:04:02 gliner | File "/usr/local/lib/python3.11/site-packages/label_studio_ml/api.py", line 126, in webhook 2024-08-15 09:04:02 gliner | model.fit(event, data) 2024-08-15 09:04:02 gliner | File "/app/model.py", line 239, in fit 2024-08-15 09:04:02 gliner | tokens, ner = self.process_training_data(task) 2024-08-15 09:04:02 gliner | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2024-08-15 09:04:02 gliner | File "/app/model.py", line 132, in process_training_data 2024-08-15 09:04:02 gliner | tokens = task['data']['tokens'] 2024-08-15 09:04:02 gliner | ~~~~^^^^^^^^^^ 2024-08-15 09:04:02 gliner | KeyError: 'tokens' 2024-08-15 09:04:02 gliner | [2024-08-15 08:04:02,089] [DEBUG] [label_studio_ml.api::log_response_info::185] Response status: 500 INTERNAL SERVER ERROR 2024-08-15 09:04:02 gliner | [2024-08-15 08:04:02,089] [DEBUG] [label_studio_ml.api::log_response_info::186] Response headers: Content-Type: text/html; charset=utf-8 2024-08-15 09:04:02 gliner | Content-Length: 265 2024-08-15 09:04:02 gliner | 2024-08-15 09:04:02 gliner | 2024-08-15 09:04:02 gliner | [2024-08-15 08:04:02,090] [DEBUG] [label_studio_ml.api::log_response_info::187] Response body: b'<!doctype html>\n\n500 Internal Server Error\n

Internal Server Error

\n

The server encountered an internal error and was unable to complete your request. Either the server is overloaded or there is an error in the application.

\n'