NeuroSpark-Instruct-1.5B

---

NeuroSpark-Instruct-Qwen2.5-1.5B-v1

A lightweight instruction-tuned language model based on Qwen2.5-1.5B, fine-tuned by QuantaSparkLabs for helpful, concise, and reliable responses.

---

🚀 Intended Use Cases

• Chatbots & assistants
• Educational tools
• Lightweight reasoning systems
• Prototyping and research
• Running on constrained hardware

---

🏋️ Training Details

• Base model: Qwen/Qwen2.5-1.5B
• Fine-tuning method: LoRA (PEFT)
• Training steps: ~507
• Epochs: 1
• Dataset size: ~50,000 instruction-response pairs
• Precision: fp16
• Optimizer: AdamW
• Training framework: Unsloth + Transformers
• Parameter count: ~1.54B

---

⚠️ Limitations

• Not trained for medical, legal, or safety-critical advice
• Knowledge cutoff inherited from the base model
• Smaller scale than multi-billion parameter models

---

🧠 System Prompt (Recommended)

For best results, use a system prompt such as:

You are NeuroSpark, a helpful, concise, and knowledgeable AI assistant.
Follow instructions carefully and provide clear, accurate responses.

---

Credits

• Qwen Team — base model
• QuantaSparkLabs — fine-tuning, dataset creation, and packaging

---

📜 License

This model is released under the Apache 2.0 License. You are free to use, modify, and distribute it, including for commercial purposes.

---

⚠️ Disclaimer

This model is provided "as-is" without warranties. The authors are not responsible for any misuse or unintended outputs.

---

🚀 Usage

🔹 Using Transformers (Recommended)

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

model_id = "QuantaSparkLabs/NeuroSpark-Instruct-1.5B"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.float16,
    device_map="auto"
)

prompt = "Explain quantum computing in simple terms."

inputs = tokenizer(prompt, return_tensors="pt").to(model.device)

outputs = model.generate(
    **inputs,
    max_new_tokens=256,
    do_sample=True,
    temperature=0.7,
    top_p=0.9
)

print(tokenizer.decode(outputs[0], skip_special_tokens=True))

---

🔹 Chat-Style Usage

messages = [
    {"role": "system", "content": "You are NeuroSpark, a helpful and concise AI assistant."},
    {"role": "user", "content": "What is artificial intelligence?"}
]

text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True
)

inputs = tokenizer(text, return_tensors="pt").to(model.device)

outputs = model.generate(
    **inputs,
    max_new_tokens=256,
    temperature=0.7
)

print(tokenizer.decode(outputs[0], skip_special_tokens=True))

---

🔹 Hardware Notes

• Runs on consumer GPUs (≈6–8 GB VRAM with fp16)
• Can run on CPU (slow, for testing only)
• Suitable for edge devices and low-resource setups

---