L'ia is a Reasoning Model trained on 500 steps in Unsloth GRPO based on 49,909 (oops) datapoints of Lambda Calculus.

This active experiment is a test to see if a model can be trained to primarily think in pure simple logic. Maybe after more time we can make the Vulcans proud!

Lia-creekside1.png

  • Developed by: Creekside
  • License: apache-2.0
  • Finetuned from model : unsloth/qwen2.5-3b-instruct-unsloth-bnb-4bit

This qwen2 model was trained 2x faster with Unsloth and Huggingface's TRL library.

Downloads last month
86
Safetensors
Model size
3.09B params
Tensor type
BF16
·
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.

Model tree for Creekside/LiA-Reasoning-Qwen-2.5-GRPO

Quantizations
1 model