65 lines
2.4 KiB
Markdown
65 lines
2.4 KiB
Markdown
|
---
|
||
|
|
language:
|
||
|
|
- id
|
||
|
|
- en
|
||
|
|
library_name: transformers
|
||
|
|
tags:
|
||
|
|
- qwen
|
||
|
|
- qwen3
|
||
|
|
- instruction-tuning
|
||
|
|
- indonesian
|
||
|
|
- alpaca
|
||
|
|
datasets:
|
||
|
|
- Ichsan2895/alpaca-gpt4-indonesian
|
||
|
|
pipeline_tag: text-generation
|
||
|
|
---
|
||
|
|
|
||
|
|
# Qwen3-4B-Indo-Alpaca
|
||
|
|
|
||
|
|
## Model Description
|
||
|
|
`Qwen3-4B-Indo-Alpaca` is an instruction-tuned language model designed for Indonesian natural language processing tasks. Built upon the Qwen3-4B base model, it has been fine-tuned using Supervised Fine-Tuning (SFT) on a translated Indonesian Alpaca-GPT4 dataset. This model is optimized to understand instructions, answer questions, and assist with general conversational tasks in Indonesian.
|
||
|
|
|
||
|
|
- **Developer:** caffeinejunkie1
|
||
|
|
- **Base Model:** Qwen3-4B
|
||
|
|
- **Language(s):** Indonesian (Primary), English
|
||
|
|
- **Model Type:** Causal Language Model (SFT)
|
||
|
|
- **License:** Apache License 2.0
|
||
|
|
|
||
|
|
## Training Data
|
||
|
|
The model was fine-tuned exclusively on the [Ichsan2895/alpaca-gpt4-indonesian](https://huggingface.co/datasets/Ichsan2895/alpaca-gpt4-indonesian) dataset. This dataset contains instruction-response pairs originally generated by GPT-4 and translated into Indonesian, providing high-quality demonstrations for instruction following.
|
||
|
|
|
||
|
|
## Intended Use
|
||
|
|
- **Primary Use Cases:** Indonesian text generation, question answering, summarization, and instruction-following.
|
||
|
|
- **Out-of-Scope:** Advanced mathematical reasoning, highly specialized medical/legal advice, or tasks requiring up-to-the-minute real-world knowledge (as the model is constrained by its training data cutoff).
|
||
|
|
|
||
|
|
## How to Use
|
||
|
|
```python
|
||
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM
|
||
|
|
|
||
|
|
model_id = "caffeinejunkie1/Qwen3-4B-Indo-Alpaca"
|
||
|
|
|
||
|
|
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
||
|
|
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")
|
||
|
|
|
||
|
|
messages = [
|
||
|
|
{"role": "system", "content": "Anda adalah asisten AI yang sangat membantu."},
|
||
|
|
{"role": "user", "content": "Jelaskan apa itu machine learning dengan bahasa yang sederhana."}
|
||
|
|
]
|
||
|
|
|
||
|
|
text = tokenizer.apply_chat_template(
|
||
|
|
messages,
|
||
|
|
tokenize=False,
|
||
|
|
add_generation_prompt=True
|
||
|
|
)
|
||
|
|
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
|
||
|
|
|
||
|
|
generated_ids = model.generate(
|
||
|
|
**model_inputs,
|
||
|
|
max_new_tokens=256
|
||
|
|
)
|
||
|
|
generated_ids = [
|
||
|
|
output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
|
||
|
|
]
|
||
|
|
|
||
|
|
response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
||
|
|
print(response)
|