OpenAIError
openai.OpenAIError (fine-tuning job cancelled: quota exceeded)
Stack trace
openai.OpenAIError: Fine-tuning job cancelled: quota exceeded. Your account has reached the maximum allowed fine-tuning quota. Please upgrade your plan or wait before retrying.
Why it happens
OpenAI enforces quota limits on fine-tuning jobs to control resource usage. When your account exceeds these limits, the fine-tuning job is automatically cancelled to prevent overuse. This can happen if you run multiple jobs or exceed your subscription's allocated fine-tuning tokens or job count.
Detection
Monitor your fine-tuning job status via the OpenAI API or dashboard and catch OpenAIError exceptions indicating quota exceeded to handle cancellations proactively.
Causes & fixes
Your OpenAI account has reached the maximum number of concurrent or total fine-tuning jobs allowed by your subscription plan.
Check your OpenAI usage dashboard and either delete old fine-tuning jobs or upgrade your subscription plan to increase your quota.
You have exceeded the total token quota allocated for fine-tuning in your billing period.
Reduce the size or number of your fine-tuning datasets or wait until your quota resets in the next billing cycle.
Multiple fine-tuning jobs are running simultaneously, exceeding concurrency limits.
Limit the number of concurrent fine-tuning jobs by queuing them or running them sequentially.
Code: broken vs fixed
from openai import OpenAI
client = OpenAI()
# This triggers quota exceeded error if limits are reached
response = client.fine_tunes.create(training_file="file-abc123", model="davinci")
print(response)import os
from openai import OpenAI, OpenAIError
client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])
try:
response = client.fine_tunes.create(training_file=os.environ["OPENAI_TRAINING_FILE_ID"], model="davinci")
print(response)
except OpenAIError as e:
if "quota exceeded" in str(e).lower():
print("Fine-tuning job cancelled due to quota exceeded. Please check your usage and upgrade your plan.")
else:
raiseWorkaround
Catch OpenAIError exceptions during fine-tuning job creation, parse the error message for quota exceeded, and implement retry logic after waiting or notify users to upgrade their plan.
Prevention
Implement usage monitoring and quota checks before submitting fine-tuning jobs, and design your system to queue jobs to stay within concurrency and token limits enforced by OpenAI.