AhıskaAI-135M-IT-v0.3
AhıskaAI-135M-IT-v0.3 is an instruction-tuned Small Language Model (~135M parameters) fine-tuned from AhıskaAI-135M-Base-v0.3. It is designed to understand multi-turn Turkish conversations, follow strict system prompt constraints, and respond in natural Turkish using the ChatML template.
Model Highlights
- Instruction Alignment: Supervised Fine-Tuned (SFT) using custom-cleaned multi-turn instruction datasets formatted in ChatML.
- Loss Masking Strategy: Trained using custom prompt masking (
labels = -100for user/system tokens), ensuring loss is calculated only on assistant responses for concise and non-hallucinating outputs. - System Prompt Support: Native support for fixed system prompts prioritizing polite, short, and accurate Turkish answers.
- Hardware-Efficient Fine-Tuning: Trained on consumer-grade hardware (NVIDIA RTX 4050 6GB GPU) using Liger Kernel acceleration (
apply_liger_kernel_to_llama) andadamw_torch_fused.
Model Details
- Base Model:
AhıskaAI/AhıskaAI-135M-Base-v0.3 - Architecture:
LlamaForCausalLMwith GQA - Parameters: ~135M
- Fine-Tuning Method: Full Parameter SFT (Supervised Fine-Tuning)
- Context Length: 512 tokens
- Template: ChatML (
<|im_start|>and<|im_end|>) - Precision:
bfloat16/float16
Supported System Prompts
The model has been optimized around two primary system personas:
"Sen kibar, sorulan soruları tam cümlelerle yanıtlayan Türkçe bir asistansın.""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-135M-IT-v0.3"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
model_id,
torch_dtype=torch.bfloat16 if torch.cuda.is_available() and torch.cuda.is_bf16_supported() else torch.float16,
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": "11 sayısından 2 çıkarırsak kaç kalır? Açıkla."}
]
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=150,
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-135M-Base-v0.3
About AhıskaAI
AhıskaAI is an independent initiative dedicated to developing efficient, high-performance Small Language Models (SLMs) tailored for the Turkish language ecosystem.