OPEA adaption of ragas (LLM-as-a-judge evaluation of Retrieval Augmented Generation)¶
OPEA’s adaption of ragas allows you to use ragas on Intel’s Gaudi AI accelerator chips.
User data¶
Please wrap your input data in datasets.Dataset
class.
from datasets import Dataset
example = {
"question": "Who is wife of Barak Obama",
"contexts": [
"Michelle Obama, wife of Barak Obama (former President of the United States of America) is an attorney",
"Barak and Michelle Obama have 2 daughters - Malia and Sasha",
],
"answer": "Michelle Obama",
"ground_truth": "Wife of Barak Obama is Michelle Obama",
}
dataset = Dataset.from_list([example])
Launch HuggingFace endpoint on Intel’s Gaudi machines¶
Please follow instructions mentioned in TGI Gaudi repo with your desired LLM such as meta-llama/Meta-Llama-3.1-70B-Instruct
.
Run OPEA ragas pipeline using your desired list of metrics¶
# note - if you wish to use answer relevancy metric, please set the embedding parameter
from langchain_community.embeddings import HuggingFaceBgeEmbeddings
embeddings = HuggingFaceBgeEmbeddings(model_name="BAAI/bge-base-en-v1.5")
from ragas import RagasMetric
ragas_metric = RagasMetric(threshold=0.5, model="<set your HF endpoint URL>", embeddings=embeddings)
print(ragas_metric.measure(dataset))
That’s it!
Troubleshooting¶
Please allow few minutes for HuggingFace endpoint to download model weights and load them. Larger models may take few more minutes. For any other issue, please file an issue and we will get back to you.