I have a dataset on Langsmith. I’m using 2 splits which combined have 8 examples.
The examples are quite big → they contain a base64
representation of a pdf, which is between 1 and 10MB each.
I build my graph locally, with langgraph dev
and then select Run experiment
and select the dataset and splits.
There’s this error in the logs:
WARNING [langsmith.client:client.py:_send_multipart_req:1969] Failed to multipart ingest runs: Connection error caused failure to POST https://api.smith.langchain.com/runs/multipart in LangSmith API. The content length of 457264450 bytes exceeds the maximum size limit of 20971520 bytes. SSLError(MaxRetryError("HTTPSConnectionPool(host='api.smith.langchain.com', port=443): Max retries exceeded with url: /runs/multipart (Caused by SSLError(SSLEOFError(8, 'EOF occurred in violation of protocol (_ssl.c:2427)')))"))
457264450 bytes = 436MB → I honestly have no idea why there’s such a big payload.
No traces are logged in the traces project, and the experiment gets marked as “completed” in the UI:
When I use single split, with 5 examples, it works.
I don’t understand why there’s 436MB in the logs, and is there a way to go around this?