Open-Assistant
Open-Assistant copied to clipboard
OpenAssistant is a chat-based assistant that understands tasks, can interact with third-party systems, and retrieve information dynamically to do so.
We need to enable Typescript strict mode to make out code safe. This needs to be incremental cause we don't want to break the build. - Set TS `strict` option...
If a user skips a task for any reason, the task shouldn't be served to that user ever again.
Add the PAQ dataset as a single turn dialogue of close-book QA, https://github.com/facebookresearch/PAQ
We need to create eval code to run against the various questions from https://huggingface.co/datasets/Hello-SimpleAI/HC3 at least a subset that we won't tain on. And we need to eval against the...
Use the prompts/story dataset from here: https://www.kaggle.com/datasets/ratthachat/writing-prompts. In addition to the prompts and story, augment with instructions such as “write a story about {prompt}, ending with the sentence {last_sentence}”. “write...
A lot of people will interact with OA. One main objective would be to keep the bot away from many biases that may originate from the base model. However, there...
Add counts of (non-deleted) messages grouped by their [lang-tag](https://github.com/LAION-AI/Open-Assistant/blob/89f9e5a88700f42351ced059c5cb1b632ccf012b/backend/oasst_backend/models/message.py#L41) to the [system stats query](https://github.com/LAION-AI/Open-Assistant/blob/89f9e5a88700f42351ced059c5cb1b632ccf012b/backend/oasst_backend/prompt_repository.py#L807). The current SystemStats response model can be found [here](https://github.com/LAION-AI/Open-Assistant/blob/89f9e5a88700f42351ced059c5cb1b632ccf012b/oasst-shared/oasst_shared/schemas/protocol.py#L384).
Establishes a setup for testing described in #84. The purpose of this PR is to enable users for writing unit tests for the backend and, probably, discord-bot components of the...
This can be solved in two steps: - [ ] Add an environment variable we can set during deployment that disables email login. - [ ] Add a ReCaptcha when...
Based on Issue #712 As discussed, I converted each dataset from a QA pair to a conversation with the agent. A couple templates are used for each dataset and the...