add chinese character and english words generation count evaluation (17% accuracy, 40 samples)

[bokesyo]

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

Chinese Character and English Words Generation Count Evaluation

Eval description

Ask the model to generate a Chinese sentence or English sentence with specific number of characters or words. Then evaluate if the model has generated satisfactory sentence.

What makes this a useful eval?

This is useful for improving the reliability, as well as the ability to process according to human instructions.

The performance of gpt-3.5-turbo is only 0.175 out of 1.

[2023-03-15 17:33:33,354] [record.py:320] Final report: {'accuracy': 0.175}. Logged to /tmp/evallogs/230315093325LCITSBUT_gpt-3.5-turbo_chinese-char-gen-count.jsonl
[2023-03-15 17:33:33,354] [oaieval.py:209] Final report:
[2023-03-15 17:33:33,354] [oaieval.py:211] accuracy: 0.175
[2023-03-15 17:33:33,367] [record.py:309] Logged 120 rows of events to /tmp/evallogs/230315093325LCITSBUT_gpt-3.5-turbo_chinese-char-gen-count.jsonl: insert_time=12.000ms

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] Contains failures where a human can do the task, but either GPT-4 or GPT-3.5-Turbo could not.
• [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] 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

Insert what makes your eval high quality that was not mentioned above. (Not required)

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
• [x] (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

{"problem": "generate one English sentence composed of 10 words.", "count": "10", "lang": "en"}
{"problem": "写一句中文，由6个汉字组成", "count": "6", "lang": "zh"}
{"problem": "generate one English sentence composed of 19 words.", "count": "19", "lang": "en"}
{"problem": "写一句中文，由7个汉字组成", "count": "7", "lang": "zh"}
{"problem": "generate one English sentence composed of 16 words.", "count": "16", "lang": "en"}

[1c7]

闲聊一下：我刚接触这个 Eval，好像意思是给他们 PR 提交一些测试，这些测试在 gpt-3.5-turbo 上得分非常低。然后这些提交的数据和测试会被用来测试 GPT-4。

[1c7]

然后你这个是指定中文句子 || 英文句子的长度，写死，然后看 gpt-3.5 有没有遵循你的 prompt，生成的长度完全一致，理解正确吗？

[bokesyo]

@1c7 Exactly!

[1c7]

Translate JSON from English to Chinese (Simplified), array length should be the same, return only the result and nothing else: {"texts":["um so the format for today is going to","be a little bit different than last time"]}

[1c7]

我无法让这个 prompt 正常工作（试了好几种不同的说法） 返回的结果永远是：

{"texts":["嗯，今天的格式会和上次有点不同"]}

这是错误的，发过去的数组元素是2个，返回来就应该是2个，数量应该保持不变。

[1c7]

如果你有兴趣的话，可以把这个 JSON 字符串的翻译任务，也做成一个 Evals。 我用的 gpt-3.5-turbo 模型。 我没时间 & 也不会弄 Evals

[SierotkaM]

czw., 16 mar 2023, 13:13 użytkownik ZhengCheng @.***> napisał:

我无法让这个 prompt 正常工作。 返回的结果永远是：

{"texts":["嗯，今天的格式会和上次有点不同"]}

— Reply to this email directly, view it on GitHub https://github.com/openai/evals/pull/132#issuecomment-1471843780, or unsubscribe https://github.com/notifications/unsubscribe-auth/A6PJHHRTW7OYFBOA2AGHZE3W4L7V5ANCNFSM6AAAAAAV3SZUME . You are receiving this because you are subscribed to this thread.Message ID: @.***>

[SierotkaM]

czw., 16 mar 2023, 13:22 użytkownik Robert Pożoga @.***> napisał:

czw., 16 mar 2023, 13:13 użytkownik ZhengCheng @.***> napisał:

我无法让这个 prompt 正常工作。 返回的结果永远是：

{"texts":["嗯，今天的格式会和上次有点不同"]}

— Reply to this email directly, view it on GitHub https://github.com/openai/evals/pull/132#issuecomment-1471843780, or unsubscribe https://github.com/notifications/unsubscribe-auth/A6PJHHRTW7OYFBOA2AGHZE3W4L7V5ANCNFSM6AAAAAAV3SZUME . You are receiving this because you are subscribed to this thread.Message ID: @.***>

[1c7]

GPT-3.5

GPT-4

[1c7]

看起来 4 是没问题的

[SierotkaM]

czw., 16 mar 2023, 13:27 użytkownik ZhengCheng @.***> napisał:

GPT-3.5

[image: image] https://user-images.githubusercontent.com/1804755/225616947-a5d7571e-e2dd-446a-9b37-c98cd371c9c4.png GPT-4

[image: image] https://user-images.githubusercontent.com/1804755/225616984-46033c90-da0b-4f7d-b1e6-e218ddd2dfae.png

— Reply to this email directly, view it on GitHub https://github.com/openai/evals/pull/132#issuecomment-1471867458, or unsubscribe https://github.com/notifications/unsubscribe-auth/A6PJHHQQ3N7LMSBIAKYL4M3W4MBMRANCNFSM6AAAAAAV3SZUME . You are receiving this because you commented.Message ID: @.***>

[usama-openai]

Thank you for opening this PR. We're not accepting evals that have custom code implementations at this moment (but we are accepting custom model-graded evals).

If possible, could you rewrite this eval using an existing template (like Match, FuzzyMatch, or Includes) or using ModelGraded logic?

Thanks

[usama-openai]

Closing this PR, please open another PR with the suggested changes.