hanaandargie/

granite-1b-base-blindspots

	Blind Spots of ibm-granite/granite-4.0-h-1b-base




	
	Model tested

Model: https://huggingface.co/ibm-granite/granite-4.0-h-1b-base

This dataset contains inputs where the model produced incorrect outputs, along with the expected output and the model's output.

!pip -q install -U transformers accelerate datasets huggingface_hub

import torch from transformers import AutoTokenizer, AutoModelForCausalLM

model_id =… See the full description on the dataset page: https://huggingface.co/datasets/hanaandargie/granite-1b-base-blindspots.