diff --git a/FaqGen/benchmark/accuracy/evaluate.py b/FaqGen/benchmark/accuracy/evaluate.py index 30998da4d..da75502ce 100644 --- a/FaqGen/benchmark/accuracy/evaluate.py +++ b/FaqGen/benchmark/accuracy/evaluate.py @@ -35,7 +35,7 @@ contexts.append([inputs_faq]) embeddings = HuggingFaceBgeEmbeddings(model_name="BAAI/bge-base-en-v1.5") -metrics_faq = ["answer_relevancy", "faithfulness", "context_utilization", "reference_free_rubrics_score"] +metrics_faq = ["answer_relevancy", "faithfulness", "context_utilization", "rubrics_score_without_reference"] metric = RagasMetric(threshold=0.5, model=llm_endpoint, embeddings=embeddings, metrics=metrics_faq) test_case = {"question": question, "answer": answer, "ground_truth": ground_truth, "contexts": contexts} diff --git a/FaqGen/benchmark/accuracy/launch_tgi.sh b/FaqGen/benchmark/accuracy/launch_tgi.sh index f4ac9eade..a504f2a41 100644 --- a/FaqGen/benchmark/accuracy/launch_tgi.sh +++ b/FaqGen/benchmark/accuracy/launch_tgi.sh @@ -11,7 +11,6 @@ docker run -it --rm \ -p $port_number:80 \ -v $volume:/data \ --runtime=habana \ - --restart always \ -e HUGGING_FACE_HUB_TOKEN=$HUGGING_FACE_HUB_TOKEN \ -e HABANA_VISIBLE_DEVICES=all \ -e OMPI_MCA_btl_vader_single_copy_mechanism=none \