wehub-resource-sync
e768098d0e
tools_continuous_delivery / Private PyPI non-main branch release (push) Has been skipped
tools_continuous_delivery / Private PyPI main branch release (push) Failing after 2m42s
Publish Promptflow Doc / Build (push) Has been cancelled
Publish Promptflow Doc / Deploy (push) Has been cancelled
Flake8 Lint / flake8 (push) Has been cancelled
Spell check CI / Spell_Check (push) Has been cancelled
30 lines
996 B
Python
30 lines
996 B
Python
from promptflow.core import tool
|
|
from typing import List
|
|
import numpy as np
|
|
|
|
|
|
def calculate_similarity(question_embedding: List, generated_question_embedding: List):
|
|
embedding1 = np.array(question_embedding)
|
|
embedding2 = np.array(generated_question_embedding)
|
|
|
|
# Compute the dot product of the two embeddings
|
|
dot_product = np.dot(embedding1, embedding2)
|
|
|
|
# Compute the L2 norms (i.e., the lengths) of each embedding
|
|
norm_embedding1 = np.linalg.norm(embedding1)
|
|
norm_embedding2 = np.linalg.norm(embedding2)
|
|
|
|
# Compute the cosine similarity
|
|
return dot_product / (norm_embedding1 * norm_embedding2)
|
|
|
|
|
|
@tool
|
|
def calculate(question_embedding: List, generated_question_embedding: List, noncommittal: bool) -> str:
|
|
cosine_sim = calculate_similarity(question_embedding, generated_question_embedding)
|
|
print("noncommittal: ")
|
|
print(noncommittal)
|
|
print(cosine_sim)
|
|
score = 5 * cosine_sim * int(not noncommittal)
|
|
|
|
return score if score >= 1 else 1
|