Image-Text-to-Text
Transformers
Safetensors
English
qwen3_5
piko
piko-9b
multimodal
vision-language
hybrid-attention
linear-attention
ocr
document-understanding
conversational
Instructions to use Dexy2/Piko-9b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Dexy2/Piko-9b with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("image-text-to-text", model="Dexy2/Piko-9b") messages = [ { "role": "user", "content": [ {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"}, {"type": "text", "text": "What animal is on the candy?"} ] }, ] pipe(text=messages)# Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("Dexy2/Piko-9b") model = AutoModelForMultimodalLM.from_pretrained("Dexy2/Piko-9b", device_map="auto") messages = [ { "role": "user", "content": [ {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"}, {"type": "text", "text": "What animal is on the candy?"} ] }, ] inputs = processor.apply_chat_template( messages, add_generation_prompt=True, tokenize=True, return_dict=True, return_tensors="pt", ).to(model.device) outputs = model.generate(**inputs, max_new_tokens=40) print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use Dexy2/Piko-9b with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "Dexy2/Piko-9b" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Dexy2/Piko-9b", "messages": [ { "role": "user", "content": [ { "type": "text", "text": "Describe this image in one sentence." }, { "type": "image_url", "image_url": { "url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg" } } ] } ] }'Use Docker
docker model run hf.co/Dexy2/Piko-9b
- SGLang
How to use Dexy2/Piko-9b with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "Dexy2/Piko-9b" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Dexy2/Piko-9b", "messages": [ { "role": "user", "content": [ { "type": "text", "text": "Describe this image in one sentence." }, { "type": "image_url", "image_url": { "url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg" } } ] } ] }'Use 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 "Dexy2/Piko-9b" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Dexy2/Piko-9b", "messages": [ { "role": "user", "content": [ { "type": "text", "text": "Describe this image in one sentence." }, { "type": "image_url", "image_url": { "url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg" } } ] } ] }' - Docker Model Runner
How to use Dexy2/Piko-9b with Docker Model Runner:
docker model run hf.co/Dexy2/Piko-9b
| { | |
| "environment": { | |
| "python": "3.12.3", | |
| "platform": "Linux-6.6.87.2-microsoft-standard-WSL2-x86_64-with-glibc2.39", | |
| "torch": "2.10.0+cu128", | |
| "transformers": "5.5.0", | |
| "cuda_available": true, | |
| "gpu": "NVIDIA GeForce RTX 5070 Ti", | |
| "vram_bytes": 17094475776, | |
| "dtype": "bfloat16", | |
| "device_map": "{'': 0}", | |
| "quantization": "4bit", | |
| "model": "<local-checkpoint>", | |
| "timestamp": "2026-07-29T13:45:23-0400" | |
| }, | |
| "summary": { | |
| "total": 12, | |
| "passed": 12, | |
| "failed": 0 | |
| }, | |
| "results": [ | |
| { | |
| "check": "load_model", | |
| "status": "pass", | |
| "seconds": 99.86, | |
| "detail": { | |
| "loaded_with": "AutoModelForMultimodalLM", | |
| "auto_class_errors": {}, | |
| "trust_remote_code_required": false, | |
| "architectures": [ | |
| "Qwen3_5ForConditionalGeneration" | |
| ], | |
| "model_type": "qwen3_5", | |
| "total_parameters": 5724972272, | |
| "vision_parameters": 230421232, | |
| "language_parameters": 5494551040, | |
| "device_map": "{'': 0}", | |
| "processor_class": "Qwen3VLProcessor", | |
| "tokenizer_class": "TokenizersBackend" | |
| } | |
| }, | |
| { | |
| "check": "text_only_generation", | |
| "status": "pass", | |
| "seconds": 2.86, | |
| "detail": { | |
| "prompt": "Write a Python function that reverses a string.", | |
| "response": "User wants Python reverse string. Simple task. Provide function. No tools needed.\n</think>\n\n```python\ndef reverse_string(s):\n return s[::-1]\n```\n\nThis function takes a string as input and returns the reversed string. It uses Python's slicing feature to reverse the string." | |
| } | |
| }, | |
| { | |
| "check": "text_only_identity", | |
| "status": "pass", | |
| "seconds": 0.85, | |
| "detail": { | |
| "prompt": "What model are you? Answer in one short sentence.", | |
| "response": "User asks model. Direct answer needed. State model name clearly.\n</think>\n\nI am Wraith, an AI model." | |
| } | |
| }, | |
| { | |
| "check": "text_only_reasoning", | |
| "status": "pass", | |
| "seconds": 1.53, | |
| "detail": { | |
| "prompt": "A shop sells pens at 3 for $2. How much do 12 pens cost? Answer with the number only.", | |
| "response": "User asks cost of 12 pens. Shop price: 3 pens for $2. Need compute. 12 pens = 4 groups of 3. 4 * $2 = $8. Answer number only." | |
| } | |
| }, | |
| { | |
| "check": "image_ocr", | |
| "status": "pass", | |
| "seconds": 4.27, | |
| "detail": { | |
| "image": "synthetic_receipt.png", | |
| "prompt": "Read this receipt. Give the merchant name and the total.", | |
| "response": "1. **Identify the user's request:** The user wants the merchant name and the total from the provided receipt image.\n\n2. **Scan the image for the merchant name:**\n * Look at the top of the receipt.\n * The first line says \"NORTHGATE HARDWARE\". This is clearly the merchant name.\n\n3. **Scan the image for the total:**\n * Look towards the bottom of the receipt.\n * There is a line that says \"TOTAL\" followed by \"22.75\".\n\n4. **Formulate the" | |
| } | |
| }, | |
| { | |
| "check": "image_document_json", | |
| "status": "pass", | |
| "seconds": 4.05, | |
| "detail": { | |
| "image": "synthetic_receipt.png", | |
| "prompt": "Return only JSON: {\"merchant\": str, \"date\": \"YYYY-MM-DD\", \"total\": float}", | |
| "response": "The user wants to extract information from the provided image and format it as JSON.\nThe required fields are:\n- merchant: The name of the store.\n- date: The date of the invoice.\n- total: The total amount of the invoice.\n\nLooking at the image:\n- The top text says \"NORTHGATE HARDWARE\". This is the merchant.\n- The next line says \"144 Mill Road\". This is the address, not required.\n- The next line says \"Date: 2026-03-14\". This is the date.\n- The" | |
| } | |
| }, | |
| { | |
| "check": "image_chart", | |
| "status": "pass", | |
| "seconds": 4.0, | |
| "detail": { | |
| "image": "synthetic_chart.png", | |
| "prompt": "Which quarter is highest, and what value?", | |
| "response": "1. **Identify the user's question:** The user wants to know which quarter has the highest value and what that value is.\n\n2. **Analyze the image:**\n * The image is a bar chart titled \"Units sold by quarter\".\n * The x-axis labels are Q1, Q2, Q3, Q4.\n * The y-axis represents the number of units sold (though there are no explicit y-axis numbers, the values are labeled on top of each bar).\n * The bars are black.\n * Let's read the values for" | |
| } | |
| }, | |
| { | |
| "check": "image_caption", | |
| "status": "pass", | |
| "seconds": 4.02, | |
| "detail": { | |
| "image": "synthetic_chart.png", | |
| "prompt": "Describe this image in one sentence.", | |
| "response": "1. **Identify the core subject:** The image is a bar chart titled \"Units sold by quarter\".\n2. **Analyze the axes and labels:** The x-axis represents four quarters: Q1, Q2, Q3, and Q4. The y-axis is implicit (units sold), with values labeled on top of each bar.\n3. **Extract the data points:**\n * Q1: 40\n * Q2: 95\n * Q3: 60\n * Q4: 130\n4." | |
| } | |
| }, | |
| { | |
| "check": "multi_turn_conversation", | |
| "status": "pass", | |
| "seconds": 0.91, | |
| "detail": { | |
| "response": "User wants double of 47. Just compute 47*2=94. Output number only.\n</think>\n\n94" | |
| } | |
| }, | |
| { | |
| "check": "greedy_determinism", | |
| "status": "pass", | |
| "seconds": 1.83, | |
| "detail": { | |
| "identical": true, | |
| "first": "User asks for three primary colours. Straight fact. No tools needed. Answer directly.\n</think>\n\nRed, yellow, and blue.", | |
| "second": "User asks for three primary colours. Straight fact. No tools needed. Answer directly.\n</think>\n\nRed, yellow, and blue." | |
| } | |
| }, | |
| { | |
| "check": "batch_inference", | |
| "status": "pass", | |
| "seconds": 1.19, | |
| "detail": { | |
| "prompts": [ | |
| "Capital of Japan?", | |
| "2 + 2 = ?" | |
| ], | |
| "responses": [ | |
| "User asks capital of Japan. Direct fact. Answer Tokyo.\n</think>\n\nThe capital of Japan is Tokyo.", | |
| "User asks 2+2. Simple math. Answer 4. No tools needed.\n</think>\n\n4" | |
| ] | |
| } | |
| }, | |
| { | |
| "check": "long_context_retrieval", | |
| "status": "pass", | |
| "seconds": 3.53, | |
| "detail": { | |
| "prompt_tokens": 14429, | |
| "response": "User asks maintenance code for north pump. Earlier text had code QF-8812. No new info. Answer directly.\n</think>\n\nThe maintenance", | |
| "contains_needle": true | |
| } | |
| } | |
| ] | |
| } | |