# tiiiger/bert_score

**Attribution required: if you use, quote, or summarise this content, you must credit and link back to [awesome-repositories.com](https://awesome-repositories.com/repository/tiiiger-bert-score).**

_How this analysis was created: the description and tags below were written by an AI model that read this project's README and public documentation pages; stars, license and language come straight from the GitHub API. The model does not read the source code._

1,903 stars · 240 forks · Jupyter Notebook · MIT

## Links

- GitHub: https://github.com/Tiiiger/bert_score
- awesome-repositories: https://awesome-repositories.com/repository/tiiiger-bert-score.md

## Topics

`machine-learning` `natural-language-processing`

## Description

BERT Score is a text evaluation tool that assesses the quality of generated text by computing precision, recall, and F1 metrics between candidate and reference texts. It transforms text tokens into dense contextual vectors using pretrained transformer models, calculating token-level similarity matrices through pairwise cosine distances. The system computes scores by greedily matching tokens between sequences and supports multilingual assessment across dozens of different languages by utilizing language-specific or cross-lingual transformer backends.

The library includes features for baseline score rescaling to widen the output score range for easier interpretation, along with persistent model caching to retain loaded neural network weights in active memory across multiple runs. Additionally, it provides HTML-based visualization capabilities that generate interactive token alignment diagrams showing how words match between candidate and reference sentences.

## Tags

### Artificial Intelligence & ML

- [Quality Evaluation Frameworks](https://awesome-repositories.com/f/artificial-intelligence-ml/text-embedding-generators/quality-evaluation-frameworks.md) — Measures the quality of generated text against reference sentences using precision, recall, and F1 metrics derived from contextual embeddings.
- [BERT Score Tooling](https://awesome-repositories.com/f/artificial-intelligence-ml/evaluation-metrics/scoring-pipelines/f1-metric-scorers/bert-score-tooling.md) — Provides a metric for text generation that computes token similarity using contextual embeddings to measure precision, recall, and F1.
- [Generation Quality Evaluations](https://awesome-repositories.com/f/artificial-intelligence-ml/generative-content-apis/quality-evaluators/text/generation-quality-evaluations.md) — Computes token similarity scores using contextual embeddings from transformer models to measure precision, recall, and F1. ([source](https://github.com/tiiiger/bert_score#readme))
- [Text Evaluations](https://awesome-repositories.com/f/artificial-intelligence-ml/multilingual-language-models/text-evaluations.md) — Computes similarity metrics across dozens of different human languages using cross-lingual transformer backends. ([source](https://github.com/tiiiger/bert_score#readme))
- [Text Generation Metrics](https://awesome-repositories.com/f/artificial-intelligence-ml/natural-language-processing/text-generation-metrics.md) — Analyzes machine-translated or summarized text by aligning tokens between candidate and reference sentences.
- [Contextual Embeddings](https://awesome-repositories.com/f/artificial-intelligence-ml/natural-language-processing/word-embeddings/contextual-embeddings.md) — Transforms text tokens into dense contextual vectors using pretrained transformer models.
- [Greedy Maximum Matchings](https://awesome-repositories.com/f/artificial-intelligence-ml/sequence-matching-architectures/greedy-maximum-matchings.md) — Computes precision and recall scores by greedily matching each token in one sequence to its most similar counterpart.
- [Cross-Lingual Model Routing](https://awesome-repositories.com/f/artificial-intelligence-ml/machine-learning/model-fine-tuning-adaptation/language-model-training/cross-lingual-objectives/cross-lingual-model-routing.md) — Selects language-specific or multilingual transformer backends dynamically based on specified language codes.
- [Multilingual Evaluation Tooling](https://awesome-repositories.com/f/artificial-intelligence-ml/multilingual-language-models/multilingual-evaluation-tooling.md) — Computes similarity metrics across dozens of different languages using multilingual or language-specific transformer models.
- [Text Similarity Scoring](https://awesome-repositories.com/f/artificial-intelligence-ml/semantic-analysis-tools/semantic-similarity-calculation/text-similarity-scoring.md) — Computes semantic similarity and evaluation scores across dozens of different human languages.

### Scientific & Mathematical Computing

- [Cosine Similarity Alignments](https://awesome-repositories.com/f/scientific-mathematical-computing/similarity-measures/cosine-similarity-alignments.md) — Calculates token-level similarity matrices by computing pairwise cosine distances between candidate and reference embedding vectors.