Downloads · 30 days
8
18% of all-time downloads
ja-t5-base-sci-cp-15k
rausch/ja-t5-base-sci-cp-15k
ja-t5-base-sci-cp-15k is a machine learning model from rausch. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for transformers.
Japanese monolingual base model continued on the SciLaD target-language split as a 15k-step control baseline.
All-time downloads
45
Public
Parameters
248M
991 MB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors990 MB · 100%
At a glance
- Library
- transformers
- Model type
- t5
- Access
- Public
- Created
- May 7, 2026
- Updated
- May 7, 2026
- SHA
- 2cb80cc8
Base models
Datasets
Model card
From the Hugging Face model README
JA-Base-CP
Japanese monolingual base model continued on the SciLaD target-language split as a 15k-step control baseline.
Model Details
This is a monolingual continued-pretraining control checkpoint reported in the paper table. It is provided for reproducibility of the baseline comparison.
- Paper name:
JA-Base-CP - Model role:
baseline-control - Source/base model: megagonlabs/t5-base-japanese-web
- Code and pipeline: GitHub repository
- Architecture: T5 encoder-decoder
- SciLaD dataset: scilons/SciLaD-all-text-v1
- Evaluation benchmark: Global-MMLU
Evaluation
Zero-shot Global-MMLU accuracy reported by the paper aggregation:
| Metric | Accuracy |
|---|---|
| Average | 22.95 |
| STEM | 21.25 |
| Humanities | 24.23 |
| Social Sciences | 21.71 |
| Other | 23.98 |
Limitations
The model is evaluated primarily with zero-shot Global-MMLU. Downstream task-specific evaluation is recommended before deployment in specialized scientific workflows.
Citation
- Title: Transferring Scientific English Pre-Trained Language Models to Multiple Languages Using Cross-Lingual Transfer
- Authors: Nikolas Rauscher, Fabio Barth, Georg Rehm
- Venue: LREC-COLING 2026, citation details TBA after publication
- Library
- transformers
- Type
- t5
- Languages
- ja
- Created
- May 7, 2026
- Updated
- May 7, 2026