NiceAsiv

Qwen3-1.7B-Base-Nuosu-LoRA

README

License: apache-2.0

Intended use / 适用范围

Chinese-to-Nuosu, Nuosu-to-Chinese, and English-related translation research.
Reproducible evaluation and continued corpus/model development.
Explicit translation instructions with deterministic decoding.

该模型适合明确指定翻译方向的研究和实验。它没有经过足够的真实多轮对话训练，不适合作为通用聊天助手，也不应被用于需要专业母语翻译质量的场景。

Training data / 训练数据

The adapter was trained through OCR-ground-truth continued pretraining, dictionary SFT, and short/long translation SFT. Sources used by this run:

NuosuBench was split into fixed train, validation, and held-out research-test partitions. The held-out partition was not used for gradient training.

Evaluation / 评测

Deterministic greedy decoding was used with max_new_tokens=96.

Table with columns: Split or target, Records, chrF2, Exact match
Split or target	Records	chrF2	Exact match
Validation, overall	192	28.1479	0.00%
Research test, overall	11,171	38.5067	17.21%
Research test, Nuosu target	4,263	25.1546	2.77%
Research test, Chinese target	2,889	39.5248	13.46%

The 192-row validation gate had no empty outputs or replacement characters and a 1.04% length-truncation rate. On the full research test, the empty-output rate was 0.009%, the replacement-character rate was 1.77%, and the length-truncation rate was 3.72%.

Automatic scores do not establish linguistic correctness. Native-speaker review remains necessary.

Known limitations / 已知问题

Some difficult Chinese-to-Nuosu prompts trigger repetitive Yi syllable loops until the generation limit.
A small number of truncated generations end with the Unicode replacement character (�).
Rare expressions, culturally specific phrases, and long translations remain unreliable.
The training distribution is translation-heavy. Open-ended writing and multi-turn dialogue behavior are not reliable.
The model has not received comprehensive native-speaker evaluation.

本模型最明显的问题是少量困难汉译彝输入会产生重复循环，直到达到生成长度上限。因此本版仅作为 research preview 发布，不应描述为成熟的彝语翻译系统。

Usage / 使用方法

python

import torchfrom peft import PeftModelfrom transformers import AutoModelForCausalLM, AutoTokenizer base_id = "Qwen/Qwen3-1.7B-Base"adapter_id = "NiceAsiv/Qwen3-1.7B-Base-Nuosu-LoRA" tokenizer = AutoTokenizer.from_pretrained(adapter_id)base = AutoModelForCausalLM.from_pretrained(    base_id,    torch_dtype=torch.bfloat16,    device_map="auto",)model = PeftModel.from_pretrained(base, adapter_id)model.eval() messages = [    {        "role": "user",        "content": "请把“欢迎来到凉山”翻译成凉山规范彝文，只输出译文。",    }]prompt = tokenizer.apply_chat_template(    messages,    tokenize=False,    add_generation_prompt=True,)inputs = tokenizer(prompt, return_tensors="pt").to(model.device)stop_ids = sorted(    {        tokenizer.eos_token_id,        tokenizer.convert_tokens_to_ids("<|im_end|>"),        tokenizer.convert_tokens_to_ids("<|endoftext|>"),    }) with torch.inference_mode():    output = model.generate(        **inputs,        max_new_tokens=96,        do_sample=False,        eos_token_id=stop_ids,        pad_token_id=tokenizer.pad_token_id,    ) print(    tokenizer.decode(        output[0, inputs["input_ids"].shape[-1]:],        skip_special_tokens=True,    ))

Use explicit translation instructions and deterministic decoding. Stop and discard an output if it enters a repetition loop.

Training summary

Base model: Qwen/Qwen3-1.7B-Base
Adapter: LoRA, rank 32, alpha 64, dropout 0.05
Precision: BF16
Training topology: three GPUs with distributed data parallelism
Stages: OCR CPT → dictionary SFT → short translation SFT → long translation SFT

Disclaimer / 免责声明

This artifact is provided for academic and research use. Users are responsible for validating generated translations and reviewing the terms of each cited data source. Do not use generated text as authoritative linguistic, educational, legal, medical, or public-service translation without qualified human review.

Available on FriendliAI

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Learn more

Container

Run this model inference with full control and performance in your environment.

Learn more

Model Details

Model Provider

NiceAsiv

Model Tree

Base

Qwen/Qwen3-1.7B-Base

Adapter

this model

Input Modalities

Text

Output Modalities