AhiskaAI-10M-Experimental-v0.1-Instruct

AhıskaAI-10M-IT-Experimental-v0.1

AhıskaAI-10M-IT-Experimental-v0.1 is an instruction-tuned micro-model (~10M parameters) fine-tuned from AhıskaAI-10M-Experimental-v0.1. It explores the boundary of instruction alignment and ChatML interaction on extreme edge-scale micro architectures.

Model Highlights

• Instruction Alignment: Supervised Fine-Tuned (SFT) using multi-turn instruction datasets formatted in ChatML.
• Loss Masking Strategy: Trained using custom prompt masking (labels = -100 for user/system tokens), ensuring loss calculation is applied strictly to assistant responses.
• System Prompt Support: Native support for fixed system prompts prioritizing polite, short, and direct answers.
• Hardware Efficiency: Extremely lightweight; runs seamlessly on basic CPU and mobile edge environments.

Model Details

• Base Model: AhıskaAI/AhıskaAI-10M-Experimental-v0.1
• Architecture: LlamaForCausalLM (Micro MHA)
• Parameters: ~10M
• Fine-Tuning Method: Full Parameter SFT
• Context Length: 512 tokens
• Vocabulary Size: 8,000
• Template: ChatML (<|im_start|> and <|im_end|>)
• Precision: float32 / float16

Supported System Prompts

The model has been optimized around two primary system personas:

1. "Sen kibar, sorulan soruları tam cümlelerle yanıtlayan Türkçe bir asistansın."
2. "Sen AhıskaAI adında Türkçe bir yapay zeka asistansın.\nGörevlerin:\n1. Sorulara doğrudan, net ve kısa cümlelerle cevap ver.\n2. Bilmediğin veya emin olmadığın konularda uydurma yapma, bilmiyorum de.\n3. Kullanıcının verdiği metin veya listeleri istenen formata sadık kalarak düzenle."

Usage (with Transformers)

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "AhıskaAI/AhıskaAI-10M-IT-Experimental-v0.1-Instruct"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
    device_map="auto"
)

# ChatML Formatting
system_prompt = "Sen kibar, sorulan soruları tam cümlelerle yanıtlayan Türkçe bir asistansın."
messages = [
    {"role": "system", "content": system_prompt},
    {"role": "user", "content": "Köpek canlı mıdır?"}
]

prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)

outputs = model.generate(
    **inputs,
    max_new_tokens=80,
    temperature=0.3,
    top_p=0.9,
    do_sample=True
)

response = tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True)
print(response)

Training Parameters & Hardware

Epochs: 1

Learning Rate: 5e-5 (Cosine Schedule)

Batch Size: 32 (effective)

Optimizer: AdamW Fused (adamw_torch_fused)

Acceleration: Liger Kernel (liger-kernel) & SDPA

Hardware: NVIDIA RTX 4050 Laptop GPU (6GB VRAM)

Related Resources

Base Model: AhıskaAI-10M-Experimental-v0.1

About AhıskaAI

AhıskaAI is an independent initiative dedicated to developing efficient, high-performance Small Language Models (SLMs) tailored for the Turkish language ecosystem.