PyPI page
Author:
Mark Marosi
Summary:
One-pass typed decisions with calibrated probabilities: decider-2b, decider-35b-a3b and the training recipe
Latest version:
1.9.0
Required dependencies:
flash-linear-attention
|
huggingface-hub
|
jinja2
|
numpy
|
torch
|
transformers
Optional dependencies:
accelerate
|
ale-py
|
datasets
|
fastapi
|
gym-super-mario-bros
|
gymnasium
|
httpx
|
imageio
|
llama-cpp-python
|
minigrid
|
mlx
|
nes-py
|
pillow
|
scikit-learn
|
uvicorn
Downloads last day:
323
Downloads last week:
2,035
Downloads last month:
6,359
7-day smoothing