Downloads · 30 days
35
5% of all-time downloads
fumetodev/Hy-MT2-1.8B-JP-Manga-Finetune-v2-GGUF
Hy-MT2-1.8B-JP-Manga-Finetune-v2-GGUF is a translation model from fumetodev. Use it when you need text moved from one language to another. The card lists the license as apache-2.0.
This repo contains a version of the Hy-MT2-1.8B model that's been finetuned further for more accurate and natural sounding English translations of conversational/manga-style Japanese than the v1 finetune. Note this im…
Downloads · 30 days
35
5% of all-time downloads
All-time downloads
724
Public
Parameters
1.8B
4.7 GB on disk
Likes
2
Public
Click a slice to open those files.
.safetensors3.6 GB · 76%
From the Hugging Face model README
This repo contains a version of the Hy-MT2-1.8B model that's been finetuned further for more accurate and natural sounding English translations of conversational/manga-style Japanese than the v1 finetune. Note this improvement likely came at the cost of translation quality for other language pairs that aren't Japanese to English.
The training data is Japanese manga dialogue, and the model is built for translating manga — short, self-contained lines, translated one line at a time.
It will still translate other Japanese, but the finetune's biases follow its training domain, and they are worth knowing before deploying it on something else:
If your input is continuous speech or long-form prose rather than discrete lines of dialogue, A/B it against stock Hy-MT2-1.8B before committing — this finetune does not dominate the base model everywhere.
The finetune was trained with a short terminology block in the prompt: a few reference translations, written in the target language, placed ahead of the Japanese text to translate. It does attend to that block — ablating it measurably lowered quality (about 1 chrF++, with some conspicuous individual failures). This is a real difference from stock Hy-MT2-1.8B, which largely ignores terminology injected into the prompt. If you have a glossary, putting it in the prompt is worthwhile here.
The instruction frame around that block was English during training; only the terminology block itself is written in the target language.
Recommended: temperature 0.15, top_k 20, top_p 0.6, repeat_penalty 1.05, min_p 0, with the sampler chain ordered penalties → top_k → top_p → temperature.
Higher temperatures measured worse. The cost is mostly variance rather than average quality, and that variance shows up as occasional invented content — so if you are seeing hallucination, lower the temperature before anything else. Note that llama.cpp and llama-server default min_p to 0.05; set it to 0 to match the conditions this model was evaluated under.