evals
evals copied to clipboard
Eval addition : Describe the meaning of Japanese onomatopoeia
Thank you for contributing an eval! โฅ๏ธ
๐จ 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. ๐จ
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. Starting April 10, the minimum eval count is 15 samples, we hope this makes it easier to create and contribute evals.
Also, pelase 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 ๐
Eval name
Describe the meaning of Japanese onomatopoeia
Eval description
The purpose of this Evals is to make you explain the meaning of Japanese onomatopoeia concisely.
What makes this a useful eval?
Better Japanese expressions can be established. We believe that onomatopoeia in Japanese is one of the most characteristic and important ways of expression in the Japanese language. Currently, when you input Japanese onomatopoeia into GPT, it suggests meanings that do not exist or ways of expression that are not used in daily life. (For example: the meaning of "zuki zuki" is described as a pulsating pain, but it is output as "the pain is felt tinglingly and gradually," etc.)
Criteria for a good eval โ
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
Basicevals or theFactModel-graded eval, or an exhaustive rubric for evaluating answers for theCriteriaModel-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.
Unique eval value
Insert what makes your eval high quality that was not mentioned above. (Not required)
Eval structure ๐๏ธ
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 ๐
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 of this form
- [x] I have used Git LFS for the Eval JSON data
- [x] (Ignore if not submitting code) I have run
pip install pre-commit; pre-commit installand have verified thatblack,isort, andautoflakeare 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": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใตใ"}], "ideal": "ๆ
ใฆใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใคใใค"}], "ideal": "็ฑใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใฃใใ"}], "ideal": "ใใคใใใชใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "ใจใใฎใใๅบๆฟ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "ใฏใฃใใใใชใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใญใชใญใช"}], "ideal": "้ญใ็ใฟ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฎใชใฎใช"}], "ideal": "้็ใซ่ฟใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใญใณใญใณ"}], "ideal": "ๅทใใฆใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใญใณใณใณใซใณใณใณ"}], "ideal": "ใใฃใคใ "}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "ๅผใฃๅผตใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฐใฆใฐใฆ"}], "ideal": "็ฉบ่
นใงใ่
นใ้ณดใ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฐใญใ"}], "ideal": "้ข็ฏใฎ้ณดใ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "ๅฐใใชๅฃฐใง็ฌใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฏใฟใฏใฟ"}], "ideal": "ใจใฆใ็ฒใใฆใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใกใใใกใ"}], "ideal": "ๅๅผ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใกใใใกใ"}], "ideal": "ๆด็ใงใใฆใใชใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฐใใฐใ"}], "ideal": "็
ฎ่พผใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใฃใใ"}], "ideal": "ๆทฑใ็ก็ "}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใฃใใ"}], "ideal": "ๅใๆใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใฉใใฉ"}], "ideal": "ใใคใใ็นฐใ่ฟใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใญใใญ"}], "ideal": "ๆฒใใฃใฆ็ถใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฐใใฐใ"}], "ideal": "ๅใ้ณดใใใชใใ้ฃฒใฟ็ฉใใฎใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฏใฉใฏใฉ"}], "ideal": "ใใพใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฐใฉใฐใฉ"}], "ideal": "ๅฎๅฎใใชใใงๆบใๅใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "ไฝๅบฆใๅใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "ไฝๅบฆใๅใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฏใณใฏใณ"}], "ideal": "ใซใใใๅ
ใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฐใณใฐใณ"}], "ideal": "ๅขใใใ้ฒ่กใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใฃใใ"}], "ideal": "ใใ่กฐใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฒใฉใฒใฉ"}], "ideal": "ๅคงๅฃฐใง็ฌใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฑใญใฑใญ"}], "ideal": "ใซใจใซใฎ้ณดใๅฃฐ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใดใฏใดใฏ"}], "ideal": "้ฃฒใฟ็ฉใ้ฃฒใ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใณใฑใณใใณใผ"}], "ideal": "ใซใใจใใฎๆณฃใๅฃฐ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "้ ใใฆไฝใใใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใดใทใดใท"}], "ideal": "ใใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใกใใใกใ"}], "ideal": "ใพใจใพใใใชใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใณใใณใ"}], "ideal": "ๅฐ้ใซ็ถใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใคใใค"}], "ideal": "่งใฐใฃใฆใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใฃใใ"}], "ideal": "ใใใใใฎ็ฉใใชใใชใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใฃใฆใ"}], "ideal": "ๅณใๆฟใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใคใ"}], "ideal": "้ใใใฎใใถใคใใ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใณใใณใ"}], "ideal": "็
ฎ่พผใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใชใใช"}], "ideal": "็ดฐใใ็ ใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใดใใดใ"}], "ideal": "ๅณใฎ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใณใญใณใญ"}], "ideal": "ๅฐใใใใฎใ่ปขใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใดใญใดใญ"}], "ideal": "ๅคงใใใใฎใ่ปขใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใณใณใณใณ"}], "ideal": "ใใใฏใฎ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใถใผใถใผ"}], "ideal": "ๆฟใใ้จ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใฃใฑใ"}], "ideal": "็ฝใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใตใฉใตใฉ"}], "ideal": "ใชใใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใถใฉใถใฉ"}], "ideal": "ๆ่งฆใใ่ใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใทใใทใ"}], "ideal": "้ใใช้จ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "ๆนฟๆฐใๅคใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใถใใใถ"}], "ideal": "ๅขใใใๆฐดใใใๅใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใทใฃใซใทใฃใซ"}], "ideal": "็ด ๆฉใๆฏใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใทใฃใญใทใฃใญ"}], "ideal": "ๆญฏใใใใฎใใ้ฃใน็ฉใ้ฃในใ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใทใฅใผ"}], "ideal": "่ใฎๅบใ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใธใฅใผใธใฅใผ"}], "ideal": "็ผใ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใ
ใใใ
ใ"}], "ideal": "็ดฐใใๆฐๆณกใ็บ็ใใใจใใฎ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใธใงใชใธใงใช"}], "ideal": "้ซญใๅใ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใธใชใธใช"}], "ideal": "ๅคช้ฝใๅผทใ็
งใไปใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "้ ๆ
ฎใชใ่ฆใคใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "ใใใๅคใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "ใใณใใใใใช็ใฟ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ในใซในใซ"}], "ideal": "้้ใใใใใใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใบใญใบใญ"}], "ideal": "่ๆใคใใใช็ใฟ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "ๆ้ทใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ในใฟในใฟ"}], "ideal": "ๆญฉใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใฃใใ"}], "ideal": "ๆฐๆใก่ฏใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใถใใถ"}], "ideal": "ๆทฑใๆฒใฟ่พผใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ในใในใ"}], "ideal": "ใชใใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "้ใใซ็ ใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ในใฉในใฉ"}], "ideal": "ๆปใใชใ้ฒใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "ใใใใคใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใบใซใบใซ"}], "ideal": "ๅใ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "ๅฏๆฐใใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใฃใใ"}], "ideal": "ไผผใฆใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใฃใจ"}], "ideal": "้ใใซไฝใใใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "้ใใซๅนใ้ขจ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใใใ"}], "ideal": "่ฝใก็ใใชใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใฃใทใ"}], "ideal": "ใใใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใ ใผใ ใผ"}], "ideal": "ไฝๅฝขใซๅใฃใฆใใชใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใ ใใ ใ"}], "ideal": "ๆฐใ็ทฉใใงใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใกใใกใ"}], "ideal": "ๅบใใใใช็ใฟ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใฃใฉใใฃใฉ"}], "ideal": "ๆตฎใคใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใฅใผใใฅใผ"}], "ideal": "ใญใใฟใฎๆณฃใๅฃฐ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใฅใณใใฅใณ"}], "ideal": "ใใใใฎๆณฃใๅฃฐ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใงใญใใงใญ"}], "ideal": "ใใตใใงๅใ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใกใใใกใใ"}], "ideal": "ๅฐใใใใฎใๅใๅใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใกใใฃใจ"}], "ideal": "็ฌ้็ใซ่ฆใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใคใผใ"}], "ideal": "้ผปใฎๅฅฅใ็ชใๅบใใใใชๅบๆฟ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใคใใค"}], "ideal": "ๅ
ๆฒข"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใคใใคใ"}], "ideal": "ใชใใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฆใใฑใ"}], "ideal": "ๆ้ใใใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฆใใฆใ"}], "ideal": "ๆญฉใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใงใใผใ"}], "ideal": "ๅนๅธ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใงใใงใ"}], "ideal": "ใ ใใใฎใชใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใซใผใณ"}], "ideal": "็็บ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใญใใญ"}], "ideal": "ๅฟ่ใฎ้ณ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใจใใจใ"}], "ideal": "ใจใใฃใใใฎ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฉใฐใฉใฐ"}], "ideal": "ๅขใใใ็ถใใฆๅบใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใญใใญ"}], "ideal": "็ฒใๆฐใฎใใ็ถๆ
"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใญใใญ"}], "ideal": "็ฒใๆฐใฎใใๆตๅ็ฉ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใณใใณ"}], "ideal": "่ปฝใๅฉใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใใณใใณ"}], "ideal": "ๅผทใๅฉใ"}
{"input": [{"role": "system", "content": "ใฆใผใถใผใๅ่ชใๅ
ฅๅใใพใใฎใงใใใฎๅ่ชใฎๆๅณใใชใในใ็ฐกๆฝใซๆใใฆใใ ใใใ"}, {"role": "user", "content": "ใฉใใใ"}], "ideal": "้่ฆใใ"}
Please fill in the INSERT_EVAL_HERE field of the PR template.
Please fill in the
INSERT_EVAL_HEREfield of the PR template.
Thanks for the heads up.
@usama-openai Thank you for your kind words. I have made the necessary corrections. Is it okay now?
@usama-openai I apologize for the repetition. Thank you for your advice. I have implemented it as you advised. What do you think?
Thanks for implementing the requested changes.
Model grading is normally used to grade the model's completion against the ideal answer or against a certain criteria. In this case, the model is supposed to grade the completion against the ideal answer, but the ideal answer is not provided to the model for grading. The model-graded spec file is closely related to closedqa, but in my opinion, the fact kind of approach would be more suitable for this case.
I apologize for this late reply.
It took me a long time to output my thoughts. I referred to fact as advised. Then I added example sentences to samples ideal answers and also changed modelgraded significantly accordingly.
I'm getting the following error while evaluating this PR:
b'File "/content/evals/evals/eval.py", line 132, in eval_sample' b'return idx, self.eval_sample(sample, rng)' b'File "/content/evals/evals/elsuite/modelgraded/classify.py", line 82, in eval_sample' b'choice, info = classify(' b'File "/content/evals/evals/elsuite/modelgraded/classify_utils.py", line 79, in classify' b'choice = get_choice(evaluation, mg.eval_type or eval_type, match_fn, choice_strings)' b'File "/content/evals/evals/elsuite/modelgraded/classify_utils.py", line 116, in get_choice' b'if eval_type.startswith("cot_classify"):' b"AttributeError: 'NoneType' object has no attribute 'startswith'"This is happening because you have not specified the
eval_typeproperty in theyamlfile.Also, the spec file you are using doesn't need
args, so you can remove themodelgraded_spec_argsproperty from theyamlfile.
I am terribly sorry for the inconvenience. We have made the change and would appreciate your confirmation.
Thanks!
You should see GPT-4 API access enabled in your account in the next few days.