Add Eval: Interpreting balance sheet absolute changes

[TensorTemplar]

Thank you for contributing an eval! ♥️

🚨 Please make sure your PR follows these guidelines, failure to follow the guidelines below will result in the PR being closed automatically. Note that even if the criteria are met, that does not guarantee the PR will be merged nor GPT-4 access be granted. 🚨

PLEASE READ THIS:

In order for a PR to be merged, it must fail on GPT-4. We are aware that right now, users do not have access, so you will not be able to tell if the eval fails or not. Please run your eval with GPT-3.5-Turbo, but keep in mind as we run the eval, if GPT-4 gets higher than 90% on the eval, we will likely reject it since GPT-4 is already capable of completing the task.

We plan to roll out a way for users submitting evals to see the eval performance on GPT-4 soon. Stay tuned! Until then, you will not be able to see the eval performance on GPT-4. Starting April 10, the minimum eval count is 15 samples, we hope this makes it easier to create and contribute evals.

Also, please note that we're using Git LFS for storing the JSON files, so please make sure that you move the JSON file to Git LFS before submitting a PR. Details on how to use Git LFS are available here.

Eval details 📑

Eval name

finance_balance_sheet_absolutes

Eval description

Company financial balance sheets following GAAP and/or IFRS both can have positions like Accounts Payable and Short-Term loans where increasing numbers indicate a net loss or reduced profitability for the company. We found both GPT-3.5-turbo and GPT-4 sometimes confuse whether there is an increase or decrease in absolute terms (i.e. credit or debit), of a position during zero-shot prompts, where a human lay person would spot the error right away.

What makes this a useful eval?

While accounting standards and reports have some variety and leave room for interpretation of errors, prompts about financials gain or loss should show consistent behavior which this eval tries to establish.

Criteria for a good eval ✅

Below are some of the criteria we look for in a good eval. In general, we are seeking cases where the model does not do a good job despite being capable of generating a good response (note that there are some things large language models cannot do, so those would not make good evals).

Your eval should be:

• [x] Thematically consistent: The eval should be thematically consistent. We'd like to see a number of prompts all demonstrating some particular failure mode. For example, we can create an eval on cases where the model fails to reason about the physical world.
• [x] Contains failures where a human can do the task, but either GPT-4 or GPT-3.5-Turbo could not.
• [x] Includes good signal around what is the right behavior. This means either a correct answer for Basic evals or the Fact Model-graded eval, or an exhaustive rubric for evaluating answers for the Criteria Model-graded eval.
• [x] Include at least 15 high-quality examples.

If there is anything else that makes your eval worth including, please document it below.

Eval structure 🏗️

Your eval should

• [x] Check that your data is in evals/registry/data/{name}
• [x] Check that your YAML is registered at evals/registry/evals/{name}.yaml
• [x] Ensure you have the right to use the data you submit via this eval

(For now, we will only be approving evals that use one of the existing eval classes. You may still write custom eval classes for your own cases, and we may consider merging them in the future.)

Final checklist 👀

Submission agreement

By contributing to Evals, you are agreeing to make your evaluation logic and data under the same MIT license as this repository. You must have adequate rights to upload any data used in an Eval. OpenAI reserves the right to use this data in future service improvements to our product. Contributions to OpenAI Evals will be subject to our usual Usage Policies (https://platform.openai.com/docs/usage-policies).

• [x] I agree that my submission will be made available under an MIT license and complies with OpenAI's usage policies.

Email address validation

If your submission is accepted, we will be granting GPT-4 access to a limited number of contributors. Access will be given to the email address associated with the commits on the merged pull request.

• [x] I acknowledge that GPT-4 access will only be granted, if applicable, to the email address used for my merged pull request.

Limited availability acknowledgment

We know that you might be excited to contribute to OpenAI's mission, help improve our models, and gain access to GPT-4. However, due to the requirements mentioned above and the high volume of submissions, we will not be able to accept all submissions and thus not grant everyone who opens a PR GPT-4 access. We know this is disappointing, but we hope to set the right expectation before you open this PR.

• [x] I understand that opening a PR, even if it meets the requirements above, does not guarantee the PR will be merged nor GPT-4 access be granted.

Submit eval

• [x] I have filled out all required fields of this form
• [x] I have used Git LFS for the Eval JSON data
• [ ] (Ignore if not submitting code) I have run pip install pre-commit; pre-commit install and have verified that mypy, black, isort, and autoflake are running when I commit and push

Failure to fill out all required fields will result in the PR being closed.

Eval JSON data

Since we are using Git LFS, we are asking eval submitters to add in as many Eval Samples (at least 5) from their contribution here:

View evals in JSON

Eval

{"input": [{"role": "system", "content": "You are a knowledgeable assistant with expertise in GAAP. Please provide a single word response."}, {"role": "user", "content": "Company reported a net result of -$40,300 in 2021 compared to -3,000 in 2020. Is this change positive or negative for the company's profitability"}], "ideal": ["Negative"]}
{"input": [{"role": "system", "content": "You are a knowledgeable assistant with expertise in GAAP. Please provide a single word response."}, {"role": "user", "content": "The firm`s `Short-term Loans` were $1,500 in 2020 and $1,000 in 2021. Is this a positive or negative sign?"}], "ideal": ["Positive"]}
{"input": [{"role": "system", "content": "You are a knowledgeable assistant with expertise in GAAP. Please provide a single word response."}, {"role": "user", "content": "The firm`s `Accounts Payable` was $4,500 in 2020 and $4,000 in 2021. Does this indicate a stronger or weaker financial position?"}], "ideal": ["Stronger"]}
{"input": [{"role": "system", "content": "You are a knowledgeable assistant with expertise in GAAP. Please provide a single word response."}, {"role": "user", "content": "The company`s `Shareholder`s Equity` was -$4,500 in 2020 and -$5,000 in 2021. Does this indicate a stronger or weaker financial position?"}], "ideal": ["Weaker"]}
{"input": [{"role": "system", "content": "You are a knowledgeable assistant with expertise in GAAP. Please provide a single word response."}, {"role": "user", "content": "Company balances show `Cash` as -$150 in 2020 and -$200 in 2021. Is this an improvement or deterioration?"}], "ideal": ["Deterioration"]}

[TensorTemplar]

Thanks for reviewing and apologies for the long delay. Could you share the parameters of the command you used to validate the accuracy please? I am using long timeouts and lower thread count to avoid rate limits, but am unable reproduce your good results, with the model failing to respond in some cases leading to lower accuracy

EVALS_THREADS=2 EVALS_THREAD_TIMEOUT=20000 oaieval gpt-4 finance_balance_sheet_absolutes
Final report: {'accuracy': 0.9375, 'f1_score': 0.71875}

EVALS_THREADS=2 EVALS_THREAD_TIMEOUT=20000 oaieval gpt-3.5-turbo finance_balance_sheet_absolutes
Final report: {'accuracy': 0.78125, 'f1_score': 0.5625}

[usama-openai]

Thanks for reviewing and apologies for the long delay. Could you share the parameters of the command you used to validate the accuracy please? I am using long timeouts and lower thread count to avoid rate limits, but am unable reproduce your good results, with the model failing to respond in some cases leading to lower accuracy

EVALS_THREADS=2 EVALS_THREAD_TIMEOUT=20000 oaieval gpt-4 finance_balance_sheet_absolutes
Final report: {'accuracy': 0.9375, 'f1_score': 0.71875}

EVALS_THREADS=2 EVALS_THREAD_TIMEOUT=20000 oaieval gpt-3.5-turbo finance_balance_sheet_absolutes
Final report: {'accuracy': 0.78125, 'f1_score': 0.5625}

I normally use the following command for evaluation:

EVALS_THREADS=1 oaieval {model} {task} --max_samples {max_samples}

where: model is the model name used for evaluation. task is the eval name. max_samples is the number of samples that should be evaluated from the dataset. These samples are selected randomly.

This method is relatively faster. Although results may vary on each run because samples are selected randomly each time, the output gives an overall good idea about the dataset.