MathBite/self_corrective_llama_3.1_8B

This is a fine-tuned version of meta-llama/Llama-3.1-8B-Instruct that has been trained to detect and mitigate hallucinations in generated text.

How to Use

Because this model uses a custom architecture, you must use trust_remote_code=True when loading it.

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "MathBite/self_corrective_llama_3.1_8B"
tokenizer = AutoTokenizer.from_pretrained(model_name)

model = AutoModelForCausalLM.from_pretrained(
    model_name,
    trust_remote_code=True
)