/LLM-Prompt-Recovery

This will be my playground to explore how to recover the LLM prompt that was used to transform a given text.

Primary LanguageJupyter Notebook

LLM-Prompt-Recovery

This will be my playground to explore how to recover the LLM prompt that was used to transform a given text.

mamba activate kllmpr

Thursday, February 29, 2024

Today I entered the Kaggle LLM Prompt Recovery competition, and I want to have a local playground for this. I have created a new LLM-Prompt-Recovery repo for this.

  1. mamba create -n kllmpr python=3.11
  2. mamba activate kllmpr
  3. mamba install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia
  4. mamba install conda-forge::transformers
  5. mamba install conda-forge::jupyterlab
  6. mamba install conda-forge::ipywidgets
  7. mamba install conda-forge::accelerate
  8. mamba install conda-forge::bitsandbytes

mamba activate kllmpr

Friday, March 1, 2024

Continue digging into the workings of gemma.

Thursday, March 7, 2024

Keep at it ...

Saturday, March 16, 2024

Attempting to run the Kaggle notebook Prompt Recovery with Gemma - KerasNLP Starter locally ... This notebook use Keras and Jax, so yeah, gonna I need to install those libraries.

Hmmm just to be safe, I am going to clone the kllmpr environment to kllmpr-2 and run the installs in that environment.

Sunday, March 17, 2024

So I didn't clone the environment yesterday, but will do it now and continue ... working through the notebook 'Local_prompt-recovery-with-gemma-kerasnlp-starter.ipynb'

  1. mamba create --name kllmpr-2 --clone kllmpr
  2. mamba activate kllmpr-2
  3. mamba install conda-forge::keras
  4. pip install --upgrade "jax[cuda11_pip]" -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html
  5. pip install --upgrade keras-cv
  6. pip install --upgrade keras-nlp
  7. pip install --upgrade keras
  8. mamba remove conda-forge::keras ... cuz keras was broken ...
  9. pip install --upgrade keras
  10. mamba install conda-forge::plotly