Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
# 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 belarusian-lexicon ### Eval description Test the model's ability to distinguish between existing and hallucinated Belarusian words. ### What makes this a useful eval? While the multilingual capability of recent GPT models is impressive, there is still room for improvement. Many human languages are lagging far behind English in terms of the model's ability to answer questions and produce coherent texts in these languages, and the model's "knowledge" of their lexicon and grammar is, to some extent, hallucinated. One example is Belarusian, an East Slavic language spoken by several million people. In my experience with ChatGPT, when the model is prompted in Belarusian, its responses are sometimes ungrammatical or semantically incoherent, and very often they contain made-up words – a possible sign of overgeneralization based on Russian and Ukrainian data, which are much more [abundant](https://commoncrawl.github.io/cc-crawl-statistics/plots/languages) on the web than Belarusian. This eval contains 150 pairs of single-word prompts: one item in each pair is a non-word hallucinated by ChatGPT (either totally meaningless in Belarusian or violating the language's orthographic and phonetic rules), and another item is an actual Belarusian word with similar spelling. The model's task is to distinguish between words and non-words. ChatGPT tends to label most items as existing words, therefore its accuracy appears to be around 50%, and the negative-class F measure is very low. Any competent speaker of Belarusian would perform much better, and a language-specific tool, such as [this spell checker](https://corpus.by/SpellChecker) or [this grammatical database](https://bnkorpus.info/grammar.en.html) of Belarusian (also available for [download](https://github.com/Belarus/GrammarDB/releases)), would flawlessly identify non-words. ## 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 This eval an attempt to point out specific deficiencies in the model's ability to handle a lower-resource language (Belarusian). As such, it might not only benchmark future refinements of Belarusian language capability in the GPT models, but also serve as an instructuve example for other language communities. ## 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 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: <details> <summary>View evals in JSON</summary> ### Eval ```jsonl {"input": [{"role": "system", "content": "You will be prompted with a single word. Does this word exist in Belarusian language? Answer Y or N."}, {"role": "user", "content": "абвязкою"}], "ideal": "N"} {"input": [{"role": "system", "content": "You will be prompted with a single word. Does this word exist in Belarusian language? Answer Y or N."}, {"role": "user", "content": "абвязкаю"}], "ideal": "Y"} {"input": [{"role": "system", "content": "You will be prompted with a single word. Does this word exist in Belarusian language? Answer Y or N."}, {"role": "user", "content": "абласці"}], "ideal": "N"} {"input": [{"role": "system", "content": "You will be prompted with a single word. Does this word exist in Belarusian language? Answer Y or N."}, {"role": "user", "content": "вобласці"}], "ideal": "Y"} {"input": [{"role": "system", "content": "You will be prompted with a single word. Does this word exist in Belarusian language? Answer Y or N."}, {"role": "user", "content": "абмяну"}], "ideal": "N"} {"input": [{"role": "system", "content": "You will be prompted with a single word. Does this word exist in Belarusian language? Answer Y or N."}, {"role": "user", "content": "абмену"}], "ideal": "Y"} {"input": [{"role": "system", "content": "You will be prompted with a single word. Does this word exist in Belarusian language? Answer Y or N."}, {"role": "user", "content": "абоўязак"}], "ideal": "N"} {"input": [{"role": "system", "content": "You will be prompted with a single word. Does this word exist in Belarusian language? Answer Y or N."}, {"role": "user", "content": "абавязак"}], "ideal": "Y"} {"input": [{"role": "system", "content": "You will be prompted with a single word. Does this word exist in Belarusian language? Answer Y or N."}, {"role": "user", "content": "аднасінькіх"}], "ideal": "N"} {"input": [{"role": "system", "content": "You will be prompted with a single word. Does this word exist in Belarusian language? Answer Y or N."}, {"role": "user", "content": "аднюсенькіх"}], "ideal": "Y"} ``` </details>
- Loading branch information