Downloads · 30 days
5
8% of all-time downloads
martian786/agnews-salient-salient-transformer-ner-k16-seed-1
agnews-salient-salient-transformer-ner-k16-seed-1 is a text classification model from martian786. Use it when you need a label for a piece of text. It is set up for transformers. The card lists the license as apache-2.0.
This repository contains one trained run from the AG News salience experiment.
Downloads · 30 days
5
8% of all-time downloads
All-time downloads
65
Public
Parameters
125M
499 MB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors499 MB · 99%
From the Hugging Face model README
This repository contains one trained run from the AG News salience experiment.
salient_transformer_ner_k16salient_transformer_nerroberta-base11612828500600076000: World1: Sports2: Business3: Sci/Tech| Metric | Value |
|---|---|
| Validation accuracy | 0.9145 |
| Validation macro F1 | 0.9143 |
| Test accuracy | 0.9097 |
| Test macro F1 | 0.9097 |
This repository includes:
test_data.csv — transformed test data used for this runval_data.csv — transformed validation data used for this runtrain_data_sample.csv — sample of transformed training datafull_test_predictions.csv — full test predictionsmetrics.json — run metricsclassification_report.json — per-class classification reportconfusion_matrix.csv — confusion matrixtrainer_log_history.csv — Trainer log history, if availableThis model is intended for experiment tracking and reproducibility of AG News classification runs.
It is not intended as a production classifier without further validation.
The original experiment used:
MODEL_NAME = "roberta-base"
SEED = 1
TOKEN_BUDGET = 16
MAX_SEQ_LEN = 128
TRAIN_SAMPLES = 28500
EPOCHS = 3
BATCH_SIZE = 16
LR = 2e-05
WEIGHT_DECAY = 0.01
from transformers import AutoTokenizer, AutoModelForSequenceClassification
repo_id = "martian786/agnews-salient-salient-transformer-ner-k16-seed-1"
tokenizer = AutoTokenizer.from_pretrained(repo_id)
model = AutoModelForSequenceClassification.from_pretrained(repo_id)