validation.py

import numpy as np
from collections import defaultdict
import tiktoken


def check_format_errors(train_dataset):
    """
    Extracted from: https://cookbook.openai.com/examples/chat_finetuning_data_prep
    """
    # Format error checks
    format_errors = defaultdict(int)

    for ex in train_dataset:
        if not isinstance(ex, dict):
            format_errors["data_type"] += 1
            continue

        messages = ex.get("messages", None)
        if not messages:
            format_errors["missing_messages_list"] += 1
            continue

        for message in messages:
            if "role" not in message or "content" not in message:
                format_errors["message_missing_key"] += 1

            if any(k not in ("role", "content", "name", "function_call", "weight") for k in message):
                format_errors["message_unrecognized_key"] += 1

            if message.get("role", None) not in ("system", "user", "assistant", "function"):
                format_errors["unrecognized_role"] += 1

            content = message.get("content", None)
            function_call = message.get("function_call", None)

            if (not content and not function_call) or not isinstance(content, str):
                format_errors["missing_content"] += 1

        if not any(message.get("role", None) == "assistant" for message in messages):
            format_errors["example_missing_assistant_message"] += 1

    if format_errors:
        print("Found errors:")
        for k, v in format_errors.items():
            print(f"{k}: {v}")
    else:
        print("No errors found")

    return format_errors if format_errors else {}

def get_distributions(train_dataset):
    """
    Extracted from: https://cookbook.openai.com/examples/chat_finetuning_data_prep

    Gets the distributions of the number of messages per example, the total number of tokens per example, and the number of assistant tokens per example.
    """
    encoding = tiktoken.get_encoding("cl100k_base")

    # not exact!
    # simplified from https://github.com/openai/openai-cookbook/blob/main/examples/How_to_count_tokens_with_tiktoken.ipynb
    def num_tokens_from_messages(messages, tokens_per_message=3, tokens_per_name=1):
        num_tokens = 0
        for message in messages:
            num_tokens += tokens_per_message
            for key, value in message.items():
                num_tokens += len(encoding.encode(value))
                if key == "name":
                    num_tokens += tokens_per_name
        num_tokens += 3
        return num_tokens

    def num_assistant_tokens_from_messages(messages):
        num_tokens = 0
        for message in messages:
            if message["role"] == "assistant":
                num_tokens += len(encoding.encode(message["content"]))
        return num_tokens


    n_missing_system = 0
    n_missing_user = 0
    n_messages = []
    convo_lens = []
    assistant_message_lens = []

    for ex in train_dataset:
        messages = ex["messages"]
        if not any(message["role"] == "system" for message in messages):
            n_missing_system += 1
        if not any(message["role"] == "user" for message in messages):
            n_missing_user += 1
        n_messages.append(len(messages))
        convo_lens.append(num_tokens_from_messages(messages))
        assistant_message_lens.append(num_assistant_tokens_from_messages(messages))
    
    return {
        "n_missing_system": n_missing_system,
        "n_missing_user": n_missing_user,
        "n_messages": n_messages,
        "convo_lens": convo_lens,
        "assistant_message_lens": assistant_message_lens
    }


def check_token_counts(train_dataset):
    """
    Extracted from: https://cookbook.openai.com/examples/chat_finetuning_data_prep
    """
    def print_distribution(values, name):
        print(f"\n#### Distribution of {name}:")
        print(f"min / max: {min(values)}, {max(values)}")
        print(f"mean / median: {np.mean(values)}, {np.median(values)}")
        print(f"p5 / p95: {np.quantile(values, 0.1)}, {np.quantile(values, 0.9)}")



    # Warnings and tokens counts
    distributions = get_distributions(train_dataset)
    n_missing_system = distributions["n_missing_system"]
    n_missing_user = distributions["n_missing_user"]
    n_messages = distributions["n_messages"]
    convo_lens = distributions["convo_lens"]
    assistant_message_lens = distributions["assistant_message_lens"]

    print("Num examples missing system message:", n_missing_system)
    print("Num examples missing user message:", n_missing_user)
    print_distribution(n_messages, "num_messages_per_example")
    print_distribution(convo_lens, "num_total_tokens_per_example")
    print_distribution(assistant_message_lens, "num_assistant_tokens_per_example")
    n_too_long = sum(l > 4096 for l in convo_lens)
    print(
        f"\n{n_too_long} examples may be over the 4096 token limit, they will be truncated during fine-tuning"
    )

    return 


def estimate_cost(train_dataset):
    """
    Extracted from: https://cookbook.openai.com/examples/chat_finetuning_data_prep
    """
    distributions = get_distributions(train_dataset)
    n_missing_system = distributions["n_missing_system"]
    n_missing_user = distributions["n_missing_user"]
    n_messages = distributions["n_messages"]
    convo_lens = distributions["convo_lens"]
    assistant_message_lens = distributions["assistant_message_lens"]



    # Pricing and default n_epochs estimate
    MAX_TOKENS_PER_EXAMPLE = 4096

    TARGET_EPOCHS = 3
    MIN_TARGET_EXAMPLES = 100
    MAX_TARGET_EXAMPLES = 25000
    MIN_DEFAULT_EPOCHS = 1
    MAX_DEFAULT_EPOCHS = 25

    n_epochs = TARGET_EPOCHS
    n_train_examples = len(train_dataset)
    if n_train_examples * TARGET_EPOCHS < MIN_TARGET_EXAMPLES:
        n_epochs = min(MAX_DEFAULT_EPOCHS, MIN_TARGET_EXAMPLES // n_train_examples)
    elif n_train_examples * TARGET_EPOCHS > MAX_TARGET_EXAMPLES:
        n_epochs = max(MIN_DEFAULT_EPOCHS, MAX_TARGET_EXAMPLES // n_train_examples)

    n_billing_tokens_in_dataset = sum(
        min(MAX_TOKENS_PER_EXAMPLE, length) for length in convo_lens
    )

    return {
        "Estimated number of tokens in dataset": n_billing_tokens_in_dataset,
        f"Estimated number of tokens that will be billed (assuming {n_epochs} training epochs)": n_epochs * n_billing_tokens_in_dataset
    }