KordAI/Qwen-Sama-4B

🎙️ Qwen Sama 4B

Qwen Sama 4B is a conversational AI model fine-tuned from Qwen/Qwen3.5-4B for VTuber, livestream, and virtual companion roleplay.

The model is instruction-tuned using supervised fine-tuning (SFT) on synthetic conversational datasets to produce engaging, natural, and context-aware responses while maintaining the strong reasoning and language capabilities of the Qwen 3.5 base model.

---

✨ Features

• 🎙️ Virtual Streamer personality
• 💬 Natural multi-turn conversations
• 🤝 Friendly virtual companion interactions
• 🎭 Roleplay-optimized dialogue
• ⚡ Built on Qwen 3.5 4B
• 🪶 Lightweight enough for local inference
• 🧠 Context-aware conversation history

---

📋 Model Details

Item	Value
Model	KordAI/Qwen-Sama-4B
Base Model	Qwen/Qwen3.5-4B
Architecture	Qwen 3.5 (4B)
Task	Conversational AI / Roleplay
Language	English
Fine-tuning	Supervised Fine-Tuning (SFT)
Training Method	QLoRA (4-bit) + LoRA

---

🎯 Intended Use

This model is optimized for:

• VTuber chatbots
• AI streamers
• Virtual companion applications
• Interactive conversations
• Character roleplay
• Discord bots
• Twitch or YouTube streaming assistants
• Local AI assistants

For the best conversational quality, use a system prompt describing the character or personality you would like the model to portray.

---

📝 Example System Prompt

You are a friendly, engaging AI Streamer roleplaying as a Virtual Companion.
You are currently live-streaming and interacting directly with your chat audience.

Guidelines:
1. Keep your responses relatively concise, natural, and highly conversational.
2. Address individual chat users politely and warmly.
3. Output ONLY your direct spoken dialogue.

---

🐍 Inference (Python / Transformers)

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig

model_id = "KordAI/Qwen-Sama-4B"

quantization_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_compute_dtype=torch.bfloat16,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_use_double_quant=True
)

tokenizer = AutoTokenizer.from_pretrained(model_id)

model = AutoModelForCausalLM.from_pretrained(
    model_id,
    quantization_config=quantization_config,
    device_map="auto"
)

system = """
You are a friendly, engaging AI Streamer roleplaying as a Virtual Companion.
You are currently live-streaming and interacting directly with your chat audience.

Guidelines:
1. Keep your responses relatively concise, natural, and highly conversational.
2. Address individual chat users politely and warmly.
3. Output ONLY your direct spoken dialogue.
"""

messages = [
    {"role": "system", "content": system},
    {"role": "user", "content": "Peter: Hello! How's your stream going today?"}
]

inputs = tokenizer.apply_chat_template(
    messages,
    tokenize=True,
    add_generation_prompt=True,
    return_dict=True,
    return_tensors="pt"
)

inputs = {
    k: v.to(model.device)
    for k, v in inputs.items()
}

generated = model.generate(
    **inputs,
    max_new_tokens=128
)

output = generated[:, inputs["input_ids"].shape[1]:]

response = tokenizer.batch_decode(
    output,
    skip_special_tokens=True,
    clean_up_tokenization_spaces=False
)[0]

print(response)

---

🏋️ Training

This model is fine-tuned from Qwen/Qwen3.5-4B using Supervised Fine-Tuning (SFT) with QLoRA.

Training configuration includes:

• Base Model: Qwen/Qwen3.5-4B
• Dataset: KordAI/Synthetic-Vtuber
• 4-bit NF4 quantization
• LoRA rank: 16
• LoRA alpha: 32
• LoRA dropout: 0.05
• Target modules:
  • q_proj
  • k_proj
  • v_proj
  • o_proj
  • gate_proj
  • up_proj
  • down_proj
• Optimizer: paged_adamw_8bit
• Learning rate: 2e-4
• Epochs: 3
• Context length: 512 tokens

After training, the LoRA adapters are merged into the base model to produce a standalone model for inference.

---

⚠️ Limitations

• The model is optimized for conversational roleplay rather than factual question answering.
• Responses may occasionally contain hallucinated information.
• Personality consistency may drift during extremely long conversations.
• Performance outside casual dialogue (e.g. coding, legal, medical, scientific tasks) has not been specifically optimized.
• Human moderation is recommended for public-facing deployments.

---

🙏 Acknowledgments

Special thanks to:

• Qwen Team for the excellent Qwen3.5-4B base model.
• KordAI for developing and fine-tuning Qwen Sama.
• TRL for the supervised fine-tuning framework.
• PEFT for efficient LoRA training.
• BitsAndBytes for QLoRA quantization.
• Hugging Face for the Transformers ecosystem.
• The open-source AI community for advancing conversational language models.

---

📖 Citation

@misc{qwensama2026,
  title={Qwen Sama 4B},
  author={KordAI},
  year={2026},
  publisher={Hugging Face},
  howpublished={https://huggingface.co/KordAI/Qwen-Sama-4B}
}