metadata
base_model: google/gemma-3-1b-it
library_name: peft
datasets:
- facebook/natural_reasoning
language:
- en
pipeline_tag: text-generation
Model Card for Model ID
fine tuned on facebook/natural_reasoning for 2.5K steps using 4bit quantization.
Evaluated using LightEval
Framework versions
- Optimizer: AdaFactor
- Bitsandbytes
- PEFT 0.14.0