Views
No views yet
Thermostatic/rosettia-chanka-4b-alpha160 (chrF++ 56.94 on held-out clean Chanka).| Metric (158-row clean Chanka held-out) | This base | Final champion | Δ |
|---|---|---|---|
| chrF++ | 43.49 | 56.94 | +13.45 |
| BLEU | 16.14 | 30.76 | +14.62 |
| token F1 | 28.94 | 46.43 | +17.49 |
| TER (↓) | 82.49 | 62.21 | −20.28 |
| Stage | Recipe |
|---|---|
| 0 | Raw unsloth/Qwen3.5-4B |
| 1 | Broad Quechua LoRA SFT (~768 steps on ~169k AmericasNLP + SomosNLP quy-spa pairs, LoRA r=64/α=128, LR 5e-6) |
| 2 | Merge the broad LoRA into the full base |
| 3 | Full-parameter FT on the clean Chanka subset (48 steps, LR=2e-6, paged_adamw_8bit) — produces this checkpoint |
1from transformers import AutoModelForCausalLM, AutoTokenizer
2import torch
3
4tok = AutoTokenizer.from_pretrained("Thermostatic/rosettia-chanka-4b-base")
5model = AutoModelForCausalLM.from_pretrained(
6 "Thermostatic/rosettia-chanka-4b-base",
7 torch_dtype=torch.bfloat16,
8 device_map="auto",
9)
10# Use the same prompt format as the champion model card.quy_Latn); not appropriate for Cuzco-Collao (quz), Bolivian (quh), or other Quechua varieties without further adaptation