You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

380 lines
15 KiB

from __future__ import annotations
import csv
import logging
import os
from typing import TYPE_CHECKING, Literal
import numpy as np
from sklearn.metrics import average_precision_score, matthews_corrcoef
from sentence_transformers.evaluation.SentenceEvaluator import SentenceEvaluator
from sentence_transformers.readers import InputExample
from sentence_transformers.similarity_functions import SimilarityFunction
from sentence_transformers.util import (
pairwise_cos_sim,
pairwise_dot_score,
pairwise_euclidean_sim,
pairwise_manhattan_sim,
)
if TYPE_CHECKING:
from sentence_transformers.SentenceTransformer import SentenceTransformer
logger = logging.getLogger(__name__)
class BinaryClassificationEvaluator(SentenceEvaluator):
"""
Evaluate a model based on the similarity of the embeddings by calculating the accuracy of identifying similar and
dissimilar sentences.
The metrics are the cosine similarity, dot score, Euclidean and Manhattan distance
The returned score is the accuracy with a specified metric.
The results are written in a CSV. If a CSV already exists, then values are appended.
The labels need to be 0 for dissimilar pairs and 1 for similar pairs.
Args:
sentences1 (List[str]): The first column of sentences.
sentences2 (List[str]): The second column of sentences.
labels (List[int]): labels[i] is the label for the pair (sentences1[i], sentences2[i]). Must be 0 or 1.
name (str, optional): Name for the output. Defaults to "".
batch_size (int, optional): Batch size used to compute embeddings. Defaults to 32.
show_progress_bar (bool, optional): If true, prints a progress bar. Defaults to False.
write_csv (bool, optional): Write results to a CSV file. Defaults to True.
truncate_dim (Optional[int], optional): The dimension to truncate sentence embeddings to. `None` uses the model's current truncation dimension. Defaults to None.
similarity_fn_names (Optional[List[Literal["cosine", "dot", "euclidean", "manhattan"]]], optional): The similarity functions to use. If not specified, defaults to the ``similarity_fn_name`` attribute of the model. Defaults to None.
Example:
::
from sentence_transformers import SentenceTransformer
from sentence_transformers.evaluation import BinaryClassificationEvaluator
from datasets import load_dataset
# Load a model
model = SentenceTransformer('all-mpnet-base-v2')
# Load a dataset with two text columns and a class label column (https://huggingface.co/datasets/sentence-transformers/quora-duplicates)
eval_dataset = load_dataset("sentence-transformers/quora-duplicates", "pair-class", split="train[-1000:]")
# Initialize the evaluator
binary_acc_evaluator = BinaryClassificationEvaluator(
sentences1=eval_dataset["sentence1"],
sentences2=eval_dataset["sentence2"],
labels=eval_dataset["label"],
name="quora_duplicates_dev",
)
results = binary_acc_evaluator(model)
'''
Binary Accuracy Evaluation of the model on the quora_duplicates_dev dataset:
Accuracy with Cosine-Similarity: 81.60 (Threshold: 0.8352)
F1 with Cosine-Similarity: 75.27 (Threshold: 0.7715)
Precision with Cosine-Similarity: 65.81
Recall with Cosine-Similarity: 87.89
Average Precision with Cosine-Similarity: 76.03
Matthews Correlation with Cosine-Similarity: 62.48
'''
print(binary_acc_evaluator.primary_metric)
# => "quora_duplicates_dev_cosine_ap"
print(results[binary_acc_evaluator.primary_metric])
# => 0.760277070888393
"""
def __init__(
self,
sentences1: list[str],
sentences2: list[str],
labels: list[int],
name: str = "",
batch_size: int = 32,
show_progress_bar: bool = False,
write_csv: bool = True,
truncate_dim: int | None = None,
similarity_fn_names: list[Literal["cosine", "dot", "euclidean", "manhattan"]] | None = None,
):
super().__init__()
self.sentences1 = sentences1
self.sentences2 = sentences2
self.labels = labels
self.truncate_dim = truncate_dim
self.similarity_fn_names = similarity_fn_names or []
assert len(self.sentences1) == len(self.sentences2)
assert len(self.sentences1) == len(self.labels)
for label in labels:
assert label == 0 or label == 1
self.write_csv = write_csv
self.name = name
self.batch_size = batch_size
if show_progress_bar is None:
show_progress_bar = (
logger.getEffectiveLevel() == logging.INFO or logger.getEffectiveLevel() == logging.DEBUG
)
self.show_progress_bar = show_progress_bar
self.csv_file = "binary_classification_evaluation" + ("_" + name if name else "") + "_results.csv"
self.csv_headers = ["epoch", "steps"]
self._append_csv_headers(self.similarity_fn_names)
def _append_csv_headers(self, similarity_fn_names: list[str]) -> None:
metrics = [
"accuracy",
"accuracy_threshold",
"f1",
"precision",
"recall",
"f1_threshold",
"ap",
"mcc",
]
for v in similarity_fn_names:
for m in metrics:
self.csv_headers.append(f"{v}_{m}")
@classmethod
def from_input_examples(cls, examples: list[InputExample], **kwargs):
sentences1 = []
sentences2 = []
scores = []
for example in examples:
sentences1.append(example.texts[0])
sentences2.append(example.texts[1])
scores.append(example.label)
return cls(sentences1, sentences2, scores, **kwargs)
def __call__(
self, model: SentenceTransformer, output_path: str | None = None, epoch: int = -1, steps: int = -1
) -> dict[str, float]:
"""
Compute the evaluation metrics for the given model.
Args:
model (SentenceTransformer): The model to evaluate.
output_path (str, optional): Path to save the evaluation results CSV file. Defaults to None.
epoch (int, optional): The epoch number. Defaults to -1.
steps (int, optional): The number of steps. Defaults to -1.
Returns:
Dict[str, float]: A dictionary containing the evaluation metrics.
"""
if epoch != -1:
if steps == -1:
out_txt = f" after epoch {epoch}"
else:
out_txt = f" in epoch {epoch} after {steps} steps"
else:
out_txt = ""
if self.truncate_dim is not None:
out_txt += f" (truncated to {self.truncate_dim})"
logger.info(f"Binary Accuracy Evaluation of the model on the {self.name} dataset{out_txt}:")
if not self.similarity_fn_names:
self.similarity_fn_names = [model.similarity_fn_name]
self._append_csv_headers(self.similarity_fn_names)
scores = self.compute_metrices(model)
file_output_data = [epoch, steps]
for header_name in self.csv_headers:
if header_name.count("_") == 1:
sim_fct, metric = header_name.split("_", maxsplit=1)
if sim_fct in scores:
file_output_data.append(scores[sim_fct][metric])
if output_path is not None and self.write_csv:
os.makedirs(output_path, exist_ok=True)
csv_path = os.path.join(output_path, self.csv_file)
if not os.path.isfile(csv_path):
with open(csv_path, newline="", mode="w", encoding="utf-8") as f:
writer = csv.writer(f)
writer.writerow(self.csv_headers)
writer.writerow(file_output_data)
else:
with open(csv_path, newline="", mode="a", encoding="utf-8") as f:
writer = csv.writer(f)
writer.writerow(file_output_data)
metrics = {
f"{short_name}_{metric}": value
for short_name, values in scores.items()
for metric, value in values.items()
}
if len(self.similarity_fn_names) > 1:
metrics.update(
{
f"max_{metric}": max(scores[short_name][metric] for short_name in scores)
for metric in scores["cosine"]
}
)
self.primary_metric = "max_ap"
else:
self.primary_metric = f"{self.similarity_fn_names[0]}_ap"
metrics = self.prefix_name_to_metrics(metrics, self.name)
self.store_metrics_in_model_card_data(model, metrics, epoch, steps)
return metrics
def compute_metrices(self, model: SentenceTransformer) -> dict[str, dict[str, float]]:
try:
# If the sentences are hashable, then we can use a set to avoid embedding the same sentences multiple
# times
sentences = list(set(self.sentences1 + self.sentences2))
except TypeError:
# Otherwise we just embed everything, e.g. if the sentences are images for evaluating a CLIP model
embeddings1 = self.embed_inputs(model, self.sentences1)
embeddings2 = self.embed_inputs(model, self.sentences2)
else:
embeddings = self.embed_inputs(model, sentences)
emb_dict = {sent: emb for sent, emb in zip(sentences, embeddings)}
embeddings1 = [emb_dict[sent] for sent in self.sentences1]
embeddings2 = [emb_dict[sent] for sent in self.sentences2]
similarity_fns = {
SimilarityFunction.COSINE.value: {
"score_fn": lambda x, y: pairwise_cos_sim(x, y),
"name": "Cosine-Similarity",
"greater_is_better": True,
},
SimilarityFunction.DOT_PRODUCT.value: {
"score_fn": lambda x, y: pairwise_dot_score(x, y),
"name": "Dot-Product",
"greater_is_better": True,
},
SimilarityFunction.MANHATTAN.value: {
"score_fn": lambda x, y: pairwise_manhattan_sim(x, y),
"name": "Manhattan-Distance",
"greater_is_better": False,
},
SimilarityFunction.EUCLIDEAN.value: {
"score_fn": lambda x, y: pairwise_euclidean_sim(x, y),
"name": "Euclidean-Distance",
"greater_is_better": False,
},
}
labels = np.asarray(self.labels)
output_scores = {}
for similarity_fn_name in self.similarity_fn_names:
similarity_fn = similarity_fns[similarity_fn_name]
scores = similarity_fn["score_fn"](embeddings1, embeddings2).detach().cpu().numpy()
greater_is_better = similarity_fn["greater_is_better"]
name = similarity_fn["name"]
acc, acc_threshold = self.find_best_acc_and_threshold(scores, labels, greater_is_better)
f1, precision, recall, f1_threshold = self.find_best_f1_and_threshold(scores, labels, greater_is_better)
ap = average_precision_score(labels, scores * (1 if greater_is_better else -1))
predicted_labels = (scores >= f1_threshold) if greater_is_better else (scores <= f1_threshold)
mcc = matthews_corrcoef(labels, predicted_labels)
logger.info(f"Accuracy with {name}: {acc * 100:.2f}\t(Threshold: {acc_threshold:.4f})")
logger.info(f"F1 with {name}: {f1 * 100:.2f}\t(Threshold: {f1_threshold:.4f})")
logger.info(f"Precision with {name}: {precision * 100:.2f}")
logger.info(f"Recall with {name}: {recall * 100:.2f}")
logger.info(f"Average Precision with {name}: {ap * 100:.2f}")
logger.info(f"Matthews Correlation with {name}: {mcc * 100:.2f}\n")
output_scores[similarity_fn_name] = {
"accuracy": acc,
"accuracy_threshold": acc_threshold,
"f1": f1,
"f1_threshold": f1_threshold,
"precision": precision,
"recall": recall,
"ap": ap,
"mcc": mcc,
}
return output_scores
def embed_inputs(
self,
model: SentenceTransformer,
sentences: str | list[str] | np.ndarray,
**kwargs,
) -> np.ndarray:
return model.encode(
sentences,
batch_size=self.batch_size,
show_progress_bar=self.show_progress_bar,
convert_to_numpy=True,
truncate_dim=self.truncate_dim,
**kwargs,
)
@staticmethod
def find_best_acc_and_threshold(scores, labels, high_score_more_similar: bool):
assert len(scores) == len(labels)
rows = list(zip(scores, labels))
rows = sorted(rows, key=lambda x: x[0], reverse=high_score_more_similar)
max_acc = 0
best_threshold = -1
positive_so_far = 0
remaining_negatives = sum(labels == 0)
for i in range(len(rows) - 1):
score, label = rows[i]
if label == 1:
positive_so_far += 1
else:
remaining_negatives -= 1
acc = (positive_so_far + remaining_negatives) / len(labels)
if acc > max_acc:
max_acc = acc
best_threshold = (rows[i][0] + rows[i + 1][0]) / 2
return max_acc, best_threshold
@staticmethod
def find_best_f1_and_threshold(scores, labels, high_score_more_similar: bool):
assert len(scores) == len(labels)
scores = np.asarray(scores)
labels = np.asarray(labels)
rows = list(zip(scores, labels))
rows = sorted(rows, key=lambda x: x[0], reverse=high_score_more_similar)
best_f1 = best_precision = best_recall = 0
threshold = 0
nextract = 0
ncorrect = 0
total_num_duplicates = sum(labels)
for i in range(len(rows) - 1):
score, label = rows[i]
nextract += 1
if label == 1:
ncorrect += 1
if ncorrect > 0:
precision = ncorrect / nextract
recall = ncorrect / total_num_duplicates
f1 = 2 * precision * recall / (precision + recall)
if f1 > best_f1:
best_f1 = f1
best_precision = precision
best_recall = recall
threshold = (rows[i][0] + rows[i + 1][0]) / 2
return best_f1, best_precision, best_recall, threshold
def get_config_dict(self):
config_dict = {}
if self.truncate_dim is not None:
config_dict["truncate_dim"] = self.truncate_dim
return config_dict