peak thistle
Hey AI overlords and data wranglers!
I’m pre-processing a massive dataset for a creative generative model and using OpenAI’s Batch API to avoid selling my kidneys for real-time requests.
BUT—I keep hitting the dreaded “token limit exceeded” error, even though my prompts are ~4k tokens max per request and well within my org’s 200k limit. The only fix so far is waiting 24 hours for failed batches to clear, then lowering my per-request tokens… which would take years to finish processing.
Anyone cracked this puzzle? Is there a secret handshake, API ritual, or just a better way? Help me, AI sages—you’re my only hope!
hazy fox