About

Tekirdağ Namık Kemal University logo ÇMF-AI-LAB logo
ÇMF-AI-LAB Academic AI Computing and Research Laboratory Tekirdağ Namık Kemal University · Çorlu Faculty of Engineering

The project and BAP support

The laboratory was established under project NKUBAP.06.GAYP.26.807, “Establishment of the Çorlu Faculty of Engineering Academic AI Computing and Research Laboratory”, funded by the Tekirdağ Namık Kemal University Scientific Research Projects (BAP) Coordination Unit.

Its aim is to let students and faculty carry out their artificial intelligence work on the university's own infrastructure: GPU-based model development and training, and the use of large language models in teaching and research. The models run on the laboratory's servers; requests and data are not sent to any AI service outside the university.

Infrastructure

  • 6 NVIDIA DGX Spark systems
  • NVIDIA GB10 (Grace Blackwell): CPU and GPU on a single chip
  • 128 GB of unified memory, shared by the CPU and the GPU (per machine)
  • 4 TB of storage (per machine)
  • Each machine has its own role: one workstation each for undergraduate students, graduate students and faculty; three machines are dedicated to language models.
  • Users sign in with their university account at a single address; their working environment opens on the machine assigned to them.

What can you do here?

Teaching

Hands-on machine learning, deep learning, computer vision and natural language processing; a separate GPU-enabled environment for every student with the libraries already installed.

Graduate studies and theses

Model training and fine-tuning (LoRA), long-running experiments, experiment tracking (TensorBoard), work with large datasets.

Research

Working with local large language models (data never leaves the university), retrieval-augmented generation (RAG), web content extraction, comparative evaluation of models.

Services

  • JupyterLab: a personal, GPU-enabled Python environment with PyTorch, Transformers and common data science libraries pre-installed; files are persistent.
  • Chat (Open WebUI): chat with the laboratory's language models, including questions about uploaded documents and images.
  • LLM API: an OpenAI-compatible API with personal keys for using the models in your own programs.
  • Model training: additional disk space for training work.
  • Remote access: use from outside the campus (faculty members).

Services are granted on application and approval, for the approved period. The services are available to members of Tekirdağ Namık Kemal University.

Language models

Open models running in the laboratory (availability depends on the user group):

  • glm-ocr
  • gemma4:e4b
  • qwen3.5:9b
  • qwen2.5-coder:14b
  • qwen2.5-coder:7b
  • qwen3-vl:8b
  • qwen3:4b
  • qwen3.8:27b
  • gemma4:31b
  • qwen3.6:35b
  • qwen3-coder:30b
  • gpt-oss:20b
  • gpt-oss:120b
  • llama3.3:70b
  • deepseek-r1:32b

Rules of use

  • The system is for education and academic research only; commercial use is not allowed.
  • Resources are shared: quotas apply, and environments idle for 30 minutes are shut down (files are kept).
  • Files and chat histories are not backed up; download important work to your own computer.
  • Do not upload personal or confidential data without permission; always verify AI output.

Privacy notice and terms of use (summary) →