README.md

metadata

model-index:
  - name: poltextlab/xlm-roberta-large-english-cap-v5
    results:
      - task:
          type: text-classification
        metrics:
          - name: Accuracy
            type: accuracy
            value: 83%
          - name: F1-Score
            type: f1
            value: 83%
tags:
  - text-classification
  - pytorch
metrics:
  - precision
  - recall
  - f1-score
language:
  - en
base_model:
  - xlm-roberta-large
pipeline_tag: text-classification
library_name: transformers
license: cc-by-4.0
extra_gated_prompt: >-
  Our models are intended for academic use only. If you are not affiliated with
  an academic institution, please provide a rationale for using our models.
  Please allow us a few business days to manually review subscriptions.
extra_gated_fields:
  Name: text
  Country: country
  Institution: text
  Institution Email: text
  Please specify your academic use case: text

xlm-roberta-large-english-cap-v5

Model description

An xlm-roberta-large model fine-tuned on english training data labeled with major topic codes from the Comparative Agendas Project.

We follow the master codebook of the Comparative Agendas Project, and all of our models use the same major topic codes.

How to use the model

from transformers import AutoTokenizer, pipeline

tokenizer = AutoTokenizer.from_pretrained("xlm-roberta-large")
pipe = pipeline(
    model="poltextlab/xlm-roberta-large-english-cap-v5",
    task="text-classification",
    tokenizer=tokenizer,
    use_fast=False,
    token="<your_hf_read_only_token>"
)

text = "<text_to_classify>"
pipe(text)

Classification Report

Overall Performance:

• Accuracy: 83%
• Macro Avg: Precision: 0.81, Recall: 0.79, F1-score: 0.80
• Weighted Avg: Precision: 0.83, Recall: 0.83, F1-score: 0.83

Per-Class Metrics:

Label	Precision	Recall	F1-score	Support
(1) Macroeconomics	0.8	0.73	0.77	11742
(2) Civil Rights	0.72	0.74	0.73	5667
(3) Health	0.88	0.9	0.89	14325
(4) Agriculture	0.88	0.85	0.87	6514
(5) Labor	0.81	0.76	0.78	7160
(6) Education	0.89	0.87	0.88	10011
(7) Environment	0.8	0.86	0.83	7385
(8) Energy	0.86	0.86	0.86	6301
(9) Immigration	0.84	0.81	0.83	2364
(10) Transportation	0.85	0.89	0.87	10769
(12) Law and Crime	0.86	0.81	0.84	16263
(13) Social Welfare	0.81	0.81	0.81	7420
(14) Housing	0.75	0.8	0.77	4982
(15) Banking, Finance, and Domestic Commerce	0.83	0.79	0.81	13510
(16) Defense	0.78	0.83	0.81	16369
(17) Technology	0.79	0.81	0.8	3952
(18) Foreign Trade	0.9	0.82	0.86	6172
(19) International Affairs	0.76	0.8	0.78	14200
(20) Government Operations	0.81	0.84	0.82	26950
(21) Public Lands	0.87	0.85	0.86	12076
(23) Culture	0.35	0.11	0.16	893
(999) No Policy Content	0.89	0.9	0.9	24532

Inference platform

This model is used by the CAP Babel Machine, an open-source and free natural language processing tool, designed to simplify and speed up projects for comparative research.

Cooperation

Model performance can be significantly improved by extending our training sets. We appreciate every submission of CAP-coded corpora (of any domain and language) at poltextlab{at}poltextlab{dot}com or by using the CAP Babel Machine.

Debugging and issues

This architecture uses the sentencepiece tokenizer. In order to run the model before transformers==4.27 you need to install it manually.