en
license: apache-2.0
task_categories:
text-generation
tags:
blind-spots
evaluation
falcon
llm
Falcon-E-1B-Base Blind Spots Dataset
Model Tested
Model: tiiuae/Falcon-E-1B-Base
Architecture: Causal decoder-only, 1.58-bit quantized base model
Parameters: ~1.8B
Developer: Technology Innovation Institute (TII)
The model was loaded on Google Colab using a free T4 GPU via the bfloat16 revision:
import torch
from transformers… See the full description on the dataset page:
https://huggingface.co/datasets/nitin966/falcon-e-1b-blind-spots.