Too much debug output

[prashbhat]

I see way too much debug output when I run langchain calls.
For example, even with verbose=False for the following RetrievalQA,

qa = RetrievalQA.from_chain_type(llm=ChatOpenAI(
model_name=chosen_model,
temperature=0,
max_tokens=num_output,
#top_p=0.3,
frequency_penalty=1,
presence_penalty=1,
),
verbose=False,
chain_type="map_reduce", retriever=retriever)

I notice that the calls to OpenAI are printed in the log. The same debug messages are printed several times - sometimes 5, 6, or 7 times.
Anyone else seen this? How do I tweak this?

Thanks.

[agola11]

It might be that the OpenAI calls are printed because we log retries as warnings (this happens on rate limits, api errors, etc). Can you provide an example of what is being printed?

[prashbhat]

Thanks for the prompt response. Here are a few examples:

---

a) Openai input (The exact same message below shows up 7 times)
DEBUG:openai:api_version=None data='{"input": [[83136, 25, 358, 1781, 264.....<Long list of integers - Trimmed>... ]], "encoding_format": "base64"}' message='Post details'

b) DEBUG message - the exact same message below shows up 7 times

DEBUG:openai:api_version=None data='{"messages": [{"role": "system", "content": "Use the following portion of a long document to see if any of the text is relevant to answer the question. \nReturn any relevant text verbatim.\n______________________\nSpeaker: So it's, it's really, it's really interesting to hear ...<Long text prompt - Trimmed>
........figure that out."}, {"role": "user", "content": "Identify the .... ....."}], "model": "gpt-4", "max_tokens": 1000, "stream": false, "n": 1, "temperature": 0.0, "frequency_penalty": 1, "presence_penalty": 1}' message='Post details'

c) DEBUG printed 7 times:

DEBUG:chromadb.db.index.hnswlib:time to run knn query: 0.00021266937255859375
time to run knn query: 0.00021266937255859375
time to run knn query: 0.00021266937255859375
time to run knn query: 0.00021266937255859375
time to run knn query: 0.00021266937255859375
time to run knn query: 0.00021266937255859375
time to run knn query: 0.00021266937255859375

d) DETAIL message printed 7 times
DETAIL:unstructured.trace:Skipping sentence because does not exceed 3 word tokens
Thanks
Skipping sentence because does not exceed 3 word tokens
Thanks
Skipping sentence because does not exceed 3 word tokens
Thanks
Skipping sentence because does not exceed 3 word tokens
Thanks
Skipping sentence because does not exceed 3 word tokens
Thanks
Skipping sentence because does not exceed 3 word tokens
Thanks
Skipping sentence because does not exceed 3 word tokens
Thanks

e) Debug message printed 7 times:
DEBUG:clickhouse_connect.driver.ctypes:Successfully import ClickHouse Connect C/Numpy optimizations
Successfully import ClickHouse Connect C/Numpy optimizations
Successfully import ClickHouse Connect C/Numpy optimizations
Successfully import ClickHouse Connect C/Numpy optimizations
Successfully import ClickHouse Connect C/Numpy optimizations
Successfully import ClickHouse Connect C/Numpy optimizations
Successfully import ClickHouse Connect C/Numpy optimizations