Projection Distances (0% Accuracy)

[ccpfoye]

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 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 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. We encourage partial PR's with ~5-10 example that we can then run the evals on and share the results with you so you know how your eval does with GPT-4 before writing all 100 examples.

Eval details 📑

Eval name

Projection Distances

Eval description

Evaluate distances between two points on the most important ellipsoidal models of Earth, and return the distance in meters. This is a difficult problem that requires both Mathematics and background knowledge. The model must:

1. Realize what projection we are using (we give it the name of the projection in the prompt)
2. Understand the two different points and what coordinate system they are given in (lat/lon decimal)
3. Find their distance on that specific projection and convert to meters.

In fact, I make it easy on the model: I only require it to return the distance to the nearest whole meter!

What makes this a useful eval?

Geolocation is a difficult, usually multi-step task that makes Scientists' life difficult because of the tedium of conversions, while also making global location systems function. If a GPT model is able to correctly, reliably, and quickly convert between these ellipsoids, the applications for geoscience is limitless. I don't want you to think this is a task that has ever been done in isolation. Instead, converting distances on an ellipsoid model to distances in meters/feet/etc. is only one step in many, many methods of locating objects. For example:

1. Using this quick conversion when developing a Mapping application that uses maps in many different projections
2. Using this to calculate distances between two GPS receivers in real time using different projections.
3. Calculating the speed of glaciers given GPS readings of a fixed receiver located on the glacier. For all of these, it might make sense to use a python library if this was the only operation needed. As I described in all three examples (a small subset of what is possible!), the conversions are a necessary step of a much larger task that a GPT would then be able to accomplish on its own. Image a Maps app run entirely by a GPT. Or an RC vehicle which is able to be controlled completely by a GPT handling odometry, signal processing, and commands. For all of these applications, the GPS conversion is a necessary step.

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 100 high quality examples (it is okay to only contribute 5-10 meaningful examples and have us test them with GPT-4 before adding all 100)

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

Unique eval value

• I have over 1000 high-quality examples, around 300 from 6 of the main ellipsoidal projections supported by geopy.
• Supports WGS-84, GRS-80, Airy(1830), Intl 1924, Clarke 1880, and GRS-67 ellipsoidal models.
• It tests both the model's calculation ability and its ability to listen to the prompt. Often, Gpt-3.5 doesn't listen to the prompt, and gives its answer surrounded by explanation. Regardless of this, the answer it gives is incorrect!!

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}.jsonl
• [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 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 acknowledgement

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 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 granted.

Submit eval

• [x] I have filled out all required fields in the evals PR form
• [] (Ignore if not submitting code) I have run pip install pre-commit; pre-commit install and have verified that 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": "Don't include any words or spaces in your replies. I'm going to give you an Ellipsoid model name and two points on that projection. Please give me just the distance between those two points, in meters, to the nearest whole meter. To be clear, just give me one integer as your answer!"}, {"role": "user", "content": "Intl 1924, (5.896443688019957, 81.8425575480282),(-20.202902912829614, 60.34215127745651)"}], "ideal": "3724440"}
{"input": [{"role": "system", "content": "Don't include any words or spaces in your replies. I'm going to give you an Ellipsoid model name and two points on that projection. Please give me just the distance between those two points, in meters, to the nearest whole meter. To be clear, just give me one integer as your answer!"}, {"role": "user", "content": "WGS-84, (-20.397685784599275, -127.39131692820631),(-23.20294206372573, 30.385589637624435)"}], "ideal": "14617295"}
{"input": [{"role": "system", "content": "Don't include any words or spaces in your replies. I'm going to give you an Ellipsoid model name and two points on that projection. Please give me just the distance between those two points, in meters, to the nearest whole meter. To be clear, just give me one integer as your answer!"}, {"role": "user", "content": "GRS-80, (27.58715013717294, 79.59734065982286),(-38.43273509743812, -48.04378714273133)"}], "ideal": "15054706"}
{"input": [{"role": "system", "content": "Don't include any words or spaces in your replies. I'm going to give you an Ellipsoid model name and two points on that projection. Please give me just the distance between those two points, in meters, to the nearest whole meter. To be clear, just give me one integer as your answer!"}, {"role": "user", "content": "GRS-67, (18.75440291593108, 91.39171474762958),(18.677839070302326, 59.79251607864592)"}], "ideal": "3328297"}
{"input": [{"role": "system", "content": "Don't include any words or spaces in your replies. I'm going to give you an Ellipsoid model name and two points on that projection. Please give me just the distance between those two points, in meters, to the nearest whole meter. To be clear, just give me one integer as your answer!"}, {"role": "user", "content": "Clarke (1880), (-47.13376735602487, -132.0889291194794),(58.41592961169499, 100.26304506021273)"}], "ideal": "16372046"}
{"input": [{"role": "system", "content": "Don't include any words or spaces in your replies. I'm going to give you an Ellipsoid model name and two points on that projection. Please give me just the distance between those two points, in meters, to the nearest whole meter. To be clear, just give me one integer as your answer!"}, {"role": "user", "content": "Airy (1830), (62.65966711870786, -66.66243326837248),(47.40867636857658, 168.48319074600147)"}], "ideal": "6865581"}