Giter VIP home page Giter VIP logo

sotopia-space's Introduction

title emoji colorFrom colorTo sdk sdk_version app_file pinned license
Sotopia Space
๐Ÿ’ป
green
pink
gradio
4.25.0
app.py
false
apache-2.0

This is a synced repository with a Huggingface Space for the Sotopia project space

Getting Started

conda create -n sotopia-space python=3.10.13; conda activate sotopia-space
python -m pip install -r requirements.txt

To run the app, run the following command:

bash start_app.sh

sotopia-space's People

Contributors

jasonqi146 avatar lwaekfjlk avatar xuhuizhou avatar algoroxyolo avatar

Stargazers

 avatar

Watchers

 avatar  avatar

sotopia-space's Issues

[FEAT]: User Interaction Logging

Description

For data and further use cases, we want to log the user's interaction with the models.

Additional Information

No response

[FEAT]: Format Prompting and Relevant Changes

Description

Sotopia-pi is trained on the same formatting guide as other sotopia models. Thus we need to add formatting guide in the end of the prompt.

Additional Information

No response

[BUG]: png cannot push to HF Space

Description of the bug

images with PNG extension name cannot be pushed to HF Space. Need to change to JPG in all history.

Steps To Reproduce

Go to Actions and push to HF Space.

Additional Information

No response

[BUG]: <clarification_hint> delete

Description of the bug

There exists special token (<clarification_hint>). Should delete them.

Steps To Reproduce

delete

Additional Information

No response

[BUG]: Model output parsing errors

Description of the bug

Model outputs parsing errors as conversation goes.

Steps To Reproduce

Talk to the model for a couple of rounds

Additional Information

No response

[BUG]: Polish the instruction

Description of the bug

Might need to point out this is a "sotopia space" instead of sotopia-pi demo.
Additionally, might need to change "model selection" to "agent model that you want to talk to"

Steps To Reproduce

change the text is fine.

Additional Information

No response

[FEAT]: Enable Scenario Selection

Description

We want to enable users to select the scenario they want the conversation to take place.

Additional Information

No response

[FEAT]: Add Parsing of HF Pipeline Output

Description

We need to use HF pipeline for custom model. However, the pipeline seems to have different output format from ChatLiteLLMs.

Additional Information

No response

[FEAT]: connect with a database

Description

We need to connect the code with a database to extract profile and save user data

Additional Information

No response

[BUG]: Sotopia-Pi models don't generate correctly

Description of the bug

Model loading or prompt formatting may be wrong in
sotopia_pi_generate

Steps To Reproduce

  1. clone this repo and install dependencies by pip install -r requirements.txt
  2. run start_app.sh
  3. Go to the uri prompted by Gradio
  4. Go the Chat tab
  5. Select cmu-lti/sotopia-pi-mistral-7b-BC_SR or cmu-lti/sotopia-pi-mistral-7b-BC_SR_4bit
  6. Talk to the model while monitoring the logs on terminal

@ProKil @XuhuiZhou @lwaekfjlk @ruiyiw

Additional Information

No response

[FEAT]: Enable Model Selection

Description

We want to optionally enable users to select the model they want to interact with.

Additional Information

No response

[BUG]: Sync to HF Space Failed

Description of the bug

The current workflows do not sync with the Huggingface Remote Space repo.

Steps To Reproduce

Whenever changes are made to main.

Additional Information

No response

[FEAT]: Hide Keys

Description

HF Spaces can only configure keys as secret env var. Need to change logics for that.

Additional Information

No response

[FEAT]: UI Style Change

Description

  1. Change Chat Tab to front
  2. Fold setting accordion except for essential info

Additional Information

No response

[FEAT]: Sotopia Prompting Logics

Description

Current prompting/template logics is approximated to sotopia rather directly adapting it. We need sotopia logics to solve these problems:

  1. We need format_bad_prompt method to make sure of the stability of response
  2. We need obtain_chain to easily extend to more LLM APIs.

Additional Information

No response

[BUG]: Fix the prompt output error

Description of the bug

Currently the UI would output the overall prompt as the output instead of just the uttearnce.

Steps To Reproduce

run UI

Additional Information

No response

[BUG]: Parsing error

Description of the bug

The word "said" should not be included in the UI.

Steps To Reproduce

image

Additional Information

No response

[BUG]: Peft model loading problem

Description of the bug

The adaptor may not be corrected loaded.

Steps To Reproduce

When generating the responses the model doesn't generate as expected.

Additional Information

No response

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.