Downloads · 30 days
8
26% of all-time downloads
sinhala-nlp/gemma-3-27b-it-TamSiPara-Ta2Si-si
gemma-3-27b-it-TamSiPara-Ta2Si-si is a translation model from sinhala-nlp. Use it when you need text moved from one language to another. It is set up for peft. The card lists the license as gemma.
A LoRA adapter for Tamil to Sinhala machine translation, trained on the TamSiPara Tamil-Sinhala parallel corpus (Pushpananda et al., 2024) as part of the SinGen Sinhala text generation benchmark.
Downloads · 30 days
8
26% of all-time downloads
All-time downloads
31
Public
Repo size
488 MB
Likes
0
Public
Click a slice to open those files.
.safetensors454 MB · 93%
From the Hugging Face model README
A LoRA adapter for Tamil to Sinhala machine translation, trained on the TamSiPara Tamil-Sinhala parallel corpus (Pushpananda et al., 2024) as part of the SinGen Sinhala text generation benchmark.
The adapter is trained on top of google/gemma-3-27b-it using the model's own chat template, with the same instruction wording as the SinGen zero-shot and few-shot prompting runs.
from transformers import AutoProcessor, AutoModelForImageTextToText
from peft import PeftModel
processor = AutoProcessor.from_pretrained("google/gemma-3-27b-it")
model = AutoModelForImageTextToText.from_pretrained("google/gemma-3-27b-it", dtype="auto", device_map="auto")
model = PeftModel.from_pretrained(model, "sinhala-nlp/gemma-3-27b-it-TamSiPara-Ta2Si-si")
(Use AutoTokenizer / AutoModelForCausalLM for text-only checkpoints.)
The model continues from the Translation: prefix inside the assistant turn.
| Training pairs | 21624 |
| Instruction language | si |
| Epochs | 3.0 |
| Effective batch size | 16 |
| Learning rate | 0.0002 |
| Max sequence length | 768 |
| LoRA r / alpha / dropout | 16 / 32 / 0.05 |
| Target modules | language-model linear layers (vision tower excluded) |
Held-out tail of TamSiPara (999 sentence pairs), whitespace-tokenized (sacreBLEU's default 13a tokenizer splits Sinhala conjuncts and vowel signs):
| Metric | Score |
|---|---|
| Corpus sacreBLEU | 16.45 |
| Sentence-level BLEU mean | 8.47 |
Derived from Gemma and governed by the Gemma Terms of Use.