hellisotherpeople/constrained-text-generation-studio

Code repo for "Most Language Models can be Poets too: An AI Writing Assistant and Constrained Text Generation Studio" at the (CAI2) workshop, jointly held at (COLING 2022)

Hi there 👋

I'm a big fan of NLP, AI, debate, and rougelike video games.

Find me on linkedin

Also, check out my webapps and models on my Huggingface Profile

Finally, I'm quite the AI artist now. See my work on Instagram

constrained-text-generation-studio's Issues

Conditioning via a learnable vector

Context: instead of a text-to-condition, I have directly a vector, which is an output from some other model (which is also learnable ).
In this context, how to do conditional fine-tuning of the GPT-2 model. From the GPT-2 model’s perspective, we can think that the dataset to train is a list of tuples of the form (condition_vec,text_sample). But note that condition_vec is an output from some other model that is also involved in the optimisation.

One naive approach is to feed this vector directly as the context vector (Assume dimensions match).
Will it work? (I could test it soon)
Or is there any better solution?

Dependence resolution for pandas when using pip install -r requirements.txt

The pandas is to be loaded with version >3, <=0.25. It is causing a recursive reinstall of each version

Collecting pandas<3,>=0.25 (from streamlit->-r requirements.txt (line 11))
  Using cached pandas-2.0.2-cp311-cp311-win_amd64.whl (10.6 MB)
  Using cached pandas-2.0.0-cp311-cp311-win_amd64.whl (11.2 MB)
  Using cached pandas-1.5.3-cp311-cp311-win_amd64.whl (10.3 MB)
  Using cached pandas-1.5.2-cp311-cp311-win_amd64.whl (10.3 MB)
  Using cached pandas-1.5.1-cp311-cp311-win_amd64.whl (10.3 MB)
  Using cached pandas-1.5.0-cp311-cp311-win_amd64.whl (10.3 MB)
  Using cached pandas-1.4.4.tar.gz (4.9 MB)
  Installing build dependencies ... done
  Getting requirements to build wheel ... done
  Preparing metadata (pyproject.toml) ... done
INFO: pip is looking at multiple versions of pandas to determine which version is compatible with other requirements. This could take a while.
  Using cached pandas-1.4.3.tar.gz (4.9 MB)
  Installing build dependencies ... done