Modelpublic

OneFly7/qwen3-embedding-0p6b-sentence-scorer-sxxx-pointwise

sourceHugging Faceupdated 3mo agoView on Hugging Face

0likes

Model Card

Qwen3 Embedding Sentence Scorer

This repository contains a trained encoder plus a regression head for sentence salience scoring.

Files

—encoder/: fine-tuned encoder saved with AutoModel.save_pretrained.
—regression_head.pt: linear regression head state dict.
—qwen3_embedding_sentence_scorer_config.json: scorer settings used during training.
—tokenizer files: tokenizer needed to encode input text.
—sentence_scorer_inference.py: minimal PyTorch inference helper.

Usage

python

from sentence_scorer_inference import load_sentence_scorer

model, tokenizer, device = load_sentence_scorer("OneFly7/qwen3-embedding-0p6b-sentence-scorer-sxxx-pointwise")
texts = ["[S001] Example article sentence. Summary candidate: example summary."]
scores = model.predict(texts, tokenizer, device=device, max_length=4096)
print(scores)

Training source directory:

text

/lustre/fswork/projects/rech/ges/uuy33zj/multi_lingual_self_distill/multi_lingual_self_distll/models/sentence_scorer_runs/qwen3_embedding_0p6b_pointwise_jz_sxxx_bs16_lr1e5

Scorer settings:

text

pooling=last_token
dropout=0.1
max_length=4096