Downloads · 30 days
84
13% of all-time downloads
FiShota/yamato-3b-v5-r128-legal-gguf
yamato-3b-v5-r128-legal-gguf is a machine learning model from FiShota. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as mit.
About the developer — One person (FiShota) building a Japanese LM stack from scratch on a single RTX 3090. Yamato is the legal/admin SFT specialist in the family. See HinoMoto-100M v15 for the from-scratch line, and h…
Downloads · 30 days
84
13% of all-time downloads
All-time downloads
634
Public
Repo size
15.6 GB
Likes
0
Public
Click a slice to open those files.
.gguf15.6 GB · 100%
From the Hugging Face model README
About the developer — One person (FiShota) building a Japanese LM stack from scratch on a single RTX 3090. Yamato is the legal/admin SFT specialist in the family. See HinoMoto-100M v15 for the from-scratch line, and hinomoto-bench-ja for the cultural-axis benchmark. GitHub · X
Best-performing variant in the rank ablation series:
| Variant | LoRA r | Overall | New domain | avg_len |
|---|---|---|---|---|
| v1 | 16 | 46.7% | 43.3% | 612 c |
| v3 | 32 | 48.9% | 43.3% | 501 c |
| v4 | 64 | 54.3% | 53.3% | 415 c |
| v5 ⭐ | 128 | 57.6% | 50.0% | 482 c |
Same 60 hand-crafted legal Q&A samples. Only LoRA rank doubled. At small-domain SFT scale, rank capacity scales much further than commonly assumed (PEFT tutorials default to r=8/16). Per-doubling deltas: +2.2 / +5.4 / +3.3.
In a sister experiment (HinoMoto multi-axis SFT, 452 samples × 4 axes ≈ 113 per axis), the same recipe at r=64 regressed by -4pt versus r=16 (96.0% → 92.0%).
→ rank scaling is bounded by samples-per-axis, not total samples.
Crossover threshold for 3B + LoRA: ~100 samples/axis (empirical, single-base observation).
→ Detailed analysis: HinoMoto 開発ノート #9 (note.com / GitHub).
yamato_legal_v02.jsonl (民法 / 刑法 / 行政手続 / 労契 / 個人情報 / 不動産登記 / 国保 / 詐欺 / 家族財産 / 年金 等)This repo includes Q3_K_M / Q4_K_M / Q5_K_M / Q6_K plus fp16. Q4_K_M recommended for general use; deployment task may benefit from per-task selection (see HinoMoto 開発ノート #8 on quantization non-monotonicity).
MIT (inherits from Sarashina2.2-3B-instruct-v0.1).