Image-Text-to-Text
Transformers
Safetensors
Turkish
English
lora
vision-language
math
exam
yks
turkish
How to use from
SGLangUse Docker images
docker run --gpus all \
--shm-size 32g \
-p 30000:30000 \
-v ~/.cache/huggingface:/root/.cache/huggingface \
--env "HF_TOKEN=<secret>" \
--ipc=host \
lmsysorg/sglang:latest \
python3 -m sglang.launch_server \
--model-path "cgalabs/yks-vlm-lora-v2" \
--host 0.0.0.0 \
--port 30000# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
-H "Content-Type: application/json" \
--data '{
"model": "cgalabs/yks-vlm-lora-v2",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'Quick Links
YKS-VLM-LoRA-v2
YKS-VLM-LoRA-v2 is a LoRA fine-tuned Vision-Language Model built on top of
Qwen2.5-VL-32B-Instruct, optimized for Turkish exam-style math questions (YKS).
This model is designed as a vision-to-structured-output component rather than a full end-to-end solver.
check us out: cga-labs.com
What this model is good at
- Reading math questions from images
- Understanding exam-style layouts (options, figures, tables)
- Producing stable, structured JSON outputs
- Acting as a preprocessing / parsing layer for downstream solvers
Typical output format:
{
"final_answer": "C"
}
Model tree for cgalabs/yks-vlm-lora-v2
Base model
Qwen/Qwen2.5-VL-32B-Instruct
Install from pip and serve model
# Install SGLang from pip: pip install sglang# Start the SGLang server: python3 -m sglang.launch_server \ --model-path "cgalabs/yks-vlm-lora-v2" \ --host 0.0.0.0 \ --port 30000# Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "cgalabs/yks-vlm-lora-v2", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'