Add pipeline tag, library name and improve model card

#1
by nielsr HF Staff - opened
Files changed (1) hide show
  1. README.md +11 -6
README.md CHANGED
@@ -1,17 +1,22 @@
1
  ---
 
 
2
  datasets:
3
  - hgissbkh/BERTJudge-Dataset
4
  language:
5
  - en
6
- base_model:
7
- - EuroBERT/EuroBERT-210m
8
  ---
 
9
  # BERTJudge-Formatted-QCR
10
 
11
  BERT-as-a-Judge is a family of encoder-based models designed for efficient, reference-based evaluation of LLM outputs. Moving beyond rigid lexical extraction and matching, these models evaluate semantic correctness, accommodating variations in phrasing and formatting while using only a fraction of the computational resources required by LLM-as-a-Judge approaches.
12
 
 
 
13
  ## Model Summary
14
- - **Paper:** [BERT-as-a-Judge: A Robust Alternative to Lexical Methods for Efficient Reference-Based LLM Evaluation](https://arxiv.org/abs/2604.09497)
15
  - **Code:** [https://github.com/artefactory/BERT-as-a-Judge](https://github.com/artefactory/BERT-as-a-Judge)
16
  - **Model Type:** Encoder-based Judge (EuroBERT-210m backbone)
17
  - **Language:** English
@@ -37,7 +42,7 @@ from bert_judge.judges import BERTJudge
37
 
38
  # 1) Initialize the judge
39
  judge = BERTJudge(
40
- model_path="artefactory/BERTJudge",
41
  trust_remote_code=True,
42
  dtype="bfloat16",
43
  )
@@ -85,10 +90,10 @@ Models follow a standardized naming structure: `BERTJudge-<Candidate_Format>-<In
85
 
86
  If you find this model useful for your research, please consider citing:
87
 
88
- ```
89
  @article{gisserotboukhlef2026bertasajudgerobustalternativelexical,
90
  title={BERT-as-a-Judge: A Robust Alternative to Lexical Methods for Efficient Reference-Based LLM Evaluation},
91
- author={Gisserot-Boukhlef, Hippolyte and Boizard, Nicolas and Malherbe, Emmanuel and Hudelot, C{\'e}line and Colombo, Pierre},
92
  year={2026},
93
  eprint={2604.09497},
94
  archivePrefix={arXiv},
 
1
  ---
2
+ base_model:
3
+ - EuroBERT/EuroBERT-210m
4
  datasets:
5
  - hgissbkh/BERTJudge-Dataset
6
  language:
7
  - en
8
+ library_name: transformers
9
+ pipeline_tag: text-classification
10
  ---
11
+
12
  # BERTJudge-Formatted-QCR
13
 
14
  BERT-as-a-Judge is a family of encoder-based models designed for efficient, reference-based evaluation of LLM outputs. Moving beyond rigid lexical extraction and matching, these models evaluate semantic correctness, accommodating variations in phrasing and formatting while using only a fraction of the computational resources required by LLM-as-a-Judge approaches.
15
 
16
+ This model was introduced in the paper [BERT-as-a-Judge: A Robust Alternative to Lexical Methods for Efficient Reference-Based LLM Evaluation](https://huggingface.co/papers/2604.09497).
17
+
18
  ## Model Summary
19
+ - **Paper:** [BERT-as-a-Judge: A Robust Alternative to Lexical Methods for Efficient Reference-Based LLM Evaluation](https://huggingface.co/papers/2604.09497)
20
  - **Code:** [https://github.com/artefactory/BERT-as-a-Judge](https://github.com/artefactory/BERT-as-a-Judge)
21
  - **Model Type:** Encoder-based Judge (EuroBERT-210m backbone)
22
  - **Language:** English
 
42
 
43
  # 1) Initialize the judge
44
  judge = BERTJudge(
45
+ model_path="hgissbkh/BERTJudge-Formatted-QCR",
46
  trust_remote_code=True,
47
  dtype="bfloat16",
48
  )
 
90
 
91
  If you find this model useful for your research, please consider citing:
92
 
93
+ ```bibtex
94
  @article{gisserotboukhlef2026bertasajudgerobustalternativelexical,
95
  title={BERT-as-a-Judge: A Robust Alternative to Lexical Methods for Efficient Reference-Based LLM Evaluation},
96
+ author={Gisserot-Boukhlef, Hippolyte and Boizard, Nicolas and Malherbe, Emmanuel and Hudelot, C{\\'e}line and Colombo, Pierre},
97
  year={2026},
98
  eprint={2604.09497},
99
  archivePrefix={arXiv},