A LLaMA 3 Youko qlora fine‑tune built on a new VNTL dataset, optimized for accurate, literal translations of Japanese visual novels to English without chat mode, using the default LLaMA 3 prompt and recommending neutral sampling (temperature 0, no repetition penalty).
Description in other languages
Statistics
Statistics as of Nov 16, 2026
Hugging Face downloads (last 30 days)
20,000
Hugging Face likes
17
Listed
Jan 21, 2025
Downloads (30-day window) over time
Show as a table
| Month | DL (30d) |
|---|---|
| Dec 2025 | 244,000 |
| Jan 2026 | 133,000 |
| Feb 2026 | 214,000 |
| Mar 2026 | 447,000 |
| Apr 2026 | 2,000,000 |
| May 2026 | 894,000 |
| Jun 2026 | 756,000 |
| Jul 2026 | 714,000 |
| Aug 2026 | 501,000 |
| Sep 2026 | 38,000 |
| Oct 2026 | 26,000 |
| Nov 2026 | 20,000 |
Downloads over the 30 days before the last snapshot of each month.
Popular in the same category
- ja-ko-vn-12b-v2-ggufhell0ks2kja‑ko‑vn‑12b V2 is a GGUF model that translates Japanese visual novel and game text into Korean, trained on CPT, SFT, and DPO with a Gemma‑3‑12B base, optimized for literal translation and supporting single‑turn inference with recommended settings.
- Hy-MT2-1.8B-JP-Manga-Finetune-v5-GGUFfumetodev1kA fine‑tuned 1.8B Hy‑MT2 Japanese→English manga translation model (v5) offering Q4_K_M GGUF and bf16 weights, trained on a cleaned manga corpus for accurate dialogue conversion in tools like FumetoReaderPlus, with strengths in short names/words but weaker real‑world proper nouns and specialist terms.
- opus-mt-ja-enHelsinki-NLP37kJapanese‑to‑English Transformer‑Align MT model from the Opus corpus, using normalization and SentencePiece preprocessing, achieves 41.7 BLEU and 0.589 chr‑F on the Tatoeba test set.
- Sugoi-14B-Ultra-GGUFsugoitoolkit10kSugoi LLM 14B Ultra (GGUF) is a Japanese‑to‑English translation model with a BLEU score of 21.38—nearly double its prior 13.67—excel at RPG‑Maker bracketed text, strong prompt adherence, and JSON output for interactive chat UIs.
- LFM2-350M-ENJP-MT-GGUFLiquidAI5kFine‑tuned, GGUF‑quantized LFM2‑350M checkpoint for near real‑time bi‑directional Japanese‑English translation of short‑to‑medium text, usable via llama.cpp.
- fugumt-ja-enstaka4kFuguMT is a Japanese‑to‑English Marian‑NMT translation model built with transformers and SentencePiece, scoring 39.1 BLEU on Tatoeba.
Resources in similar research fields
plamo-translate-clipfnet364A command-line interface for translation using the plamo-2-translate model with local execution.GitHubShared: Machine Translation- fugumt-en-jastaka4kFuguMT is a Marian‑NMT based English‑to‑Japanese translation model built with Hugging Face Transformers and SentencePiece, achieving a BLEU score of 32.7 on Tatoeba.Shared: Machine Translation
- plamo-2-translatepfnet2kPLaMo Translation Model is a large‑scale language model created by Preferred Networks for translation tasks, available in base, post‑trained, and evaluation variants, released under the PLaMo community license and not instruction‑tuned for chat or other downstream uses.Shared: Machine Translation
VISAku-nlp14An ambiguous subtitles dataset for visual scene-aware machine translationGitHubShared: Machine Translation
open-mantra-datasetmantra-inc208Dataset introduced in the paper Towards Fully Automated Manga Translation presented in AAAI21GitHubShared: Machine Translation
koharukoharu-rs6.7kAutomated manga translation tool with LLM, written in Rust.GitHubShared: Machine Translation