Too many retries trying to get the correct output format. Try simplifying the requirements.', {'subquestions': "ValueError('json output should start and end with { and }')"})

[javanan]

backend-1 | 2025-02-20 02:34:46,913 - app.rag.chat:215 - ERROR - ('Too many retries trying to get the correct output format. Try simplifying the requirements.', {'subquestions': "ValueError('json output should start and end with { and }')"})
backend-1 | Traceback (most recent call last):
backend-1 | File "/app/app/rag/chat.py", line 212, in chat
backend-1 | for event in self._chat():
backend-1 | File "/app/app/rag/chat.py", line 700, in _chat
backend-1 | ) = yield from self._search_kg(
backend-1 | ^^^^^^^^^^^^^^^^^^^^^^^^^^^
backend-1 | File "/app/app/rag/chat.py", line 278, in _search_kg
backend-1 | sub_queries = graph_index.intent_analyze(
backend-1 | ^^^^^^^^^^^^^^^^^^^^^^^^^^^
backend-1 | File "/app/app/rag/knowledge_graph/base.py", line 237, in intent_analyze
backend-1 | intents = self._intents.analyze(chat_content)
backend-1 | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
backend-1 | File "/app/app/rag/knowledge_graph/intent.py", line 77, in analyze
backend-1 | return self.intent_anlysis_prog(query=query).subquestions
backend-1 | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
backend-1 | File "/usr/local/lib/python3.11/site-packages/dspy/primitives/program.py", line 26, in call
backend-1 | return self.forward(*args, **kwargs)
backend-1 | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
backend-1 | File "/app/app/rag/knowledge_graph/intent.py", line 67, in forward
backend-1 | return self.prog(query=query)
backend-1 | ^^^^^^^^^^^^^^^^^^^^^^
backend-1 | File "/usr/local/lib/python3.11/site-packages/dspy/primitives/program.py", line 26, in call
backend-1 | return self.forward(*args, **kwargs)
backend-1 | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
backend-1 | File "/usr/local/lib/python3.11/site-packages/dspy/functional/functional.py", line 369, in forward
backend-1 | raise ValueError(
backend-1 | ValueError: ('Too many retries trying to get the correct output format. Try simplifying the requirements.', {'subquestions': "ValueError('json output should start and end with { and }')"})

Not finished
Sources
Empty
Answer
This chatbot is in Beta.
All generated information should be verified prior to use.
Failed to generate response
Encountered an error while processing the chat. Please try again later.

[Mini256]

Which model are you currently using?

[leeyuxxuan]

I have encountered the same problem. May I ask if there is any solution to it?

[Mini256]

@javanan @leeyuxxuan Thanks for your feedback! I am trying to solve this kind of compatibility issue. Could you mind tell me the llm model you used? I may add it to the integration tests.

For workaround:

Maybe you can add some few-shot exampels for this module by specifing the following environment vairbale in the .env file to resolve it:

COMPLIED_INTENT_ANALYSIS_PROGRAM_PATH=./dspy_compiled_program/decompose_query_program