OpenAI fine-tuning API in Prompt Engineering / GenAI - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is the purpose of the OpenAI fine-tuning API?

The OpenAI fine-tuning API allows users to customize a base language model by training it on their own specific data, improving performance on tasks relevant to their needs.

beginner

What type of data format is required for fine-tuning with the OpenAI API?

The data must be in JSONL format, where each line is a JSON object with at least two keys: "prompt" and "completion", representing input and desired output respectively.

intermediate

Which OpenAI model is commonly used as a base for fine-tuning?

Models like "davinci" or "curie" are commonly used as base models for fine-tuning because they provide strong general language understanding.

intermediate

What are the main steps to fine-tune a model using the OpenAI fine-tuning API?

1. Prepare and format your training data in JSONL.<br>2. Upload the data file to OpenAI.<br>3. Create a fine-tuning job specifying the base model and training file.<br>4. Monitor the training process.<br>5. Use the fine-tuned model for predictions.

intermediate

How can you evaluate the performance of a fine-tuned OpenAI model?

You can evaluate it by testing the model on a set of prompts and comparing the outputs to expected completions, measuring metrics like accuracy, relevance, or loss depending on the task.

What file format must training data be in for OpenAI fine-tuning?

A TXT

B CSV

C XML

D JSONL

OpenAI fine-tuning requires data in JSONL format, where each line is a JSON object with prompt and completion.

Which key is NOT required in each JSONL training example?

A "prompt"

B "label"

C "completion"

D Both A and B are required

Only "prompt" and "completion" keys are required; "label" is not used in OpenAI fine-tuning data.

What is the first step in the OpenAI fine-tuning process?

A Prepare training data

B Create fine-tuning job

C Upload training data

D Use the fine-tuned model

You must first prepare and format your training data before uploading or starting fine-tuning.

Which OpenAI model is commonly used as a base for fine-tuning?

A GPT-2

B ResNet

C davinci

D BERT

"davinci" is a powerful OpenAI model often used as a base for fine-tuning.

How do you monitor the progress of a fine-tuning job?

A Check training logs via API or dashboard

B Wait silently until completion

C Use a separate monitoring tool unrelated to OpenAI

D Fine-tuning jobs do not provide progress info

OpenAI provides logs and status updates via API or dashboard to monitor fine-tuning progress.

Explain the process of fine-tuning a model using the OpenAI fine-tuning API.

Think about the steps from data preparation to using the model.

Describe how you would evaluate if your fine-tuned OpenAI model is working well.

Consider how to check if the model learned what you wanted.