Instructions to use leilei535406/SmolVLM2-256M-Video-Instruct-vqav2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use leilei535406/SmolVLM2-256M-Video-Instruct-vqav2 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="leilei535406/SmolVLM2-256M-Video-Instruct-vqav2")

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("leilei535406/SmolVLM2-256M-Video-Instruct-vqav2")
model = AutoModelForMultimodalLM.from_pretrained("leilei535406/SmolVLM2-256M-Video-Instruct-vqav2", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use leilei535406/SmolVLM2-256M-Video-Instruct-vqav2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "leilei535406/SmolVLM2-256M-Video-Instruct-vqav2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "leilei535406/SmolVLM2-256M-Video-Instruct-vqav2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/leilei535406/SmolVLM2-256M-Video-Instruct-vqav2

SGLang

How to use leilei535406/SmolVLM2-256M-Video-Instruct-vqav2 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 "leilei535406/SmolVLM2-256M-Video-Instruct-vqav2" \
    --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": "leilei535406/SmolVLM2-256M-Video-Instruct-vqav2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "leilei535406/SmolVLM2-256M-Video-Instruct-vqav2" \
        --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": "leilei535406/SmolVLM2-256M-Video-Instruct-vqav2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use leilei535406/SmolVLM2-256M-Video-Instruct-vqav2 with Docker Model Runner:
```
docker model run hf.co/leilei535406/SmolVLM2-256M-Video-Instruct-vqav2
```

SmolVLM2-256M-Video-Instruct-vqav2

This model is a fine-tuned version of HuggingFaceTB/SmolVLM2-256M-Video-Instruct on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.5032

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 2e-05
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 8
total_train_batch_size: 64
optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
4.5983	0.2153	20	4.0399
2.2296	0.4307	40	1.9635
1.543	0.6460	60	1.4162
1.1647	0.8614	80	1.0928
0.9353	1.0754	100	0.8797
0.7645	1.2907	120	0.7375
0.6545	1.5061	140	0.6353
0.5837	1.7214	160	0.5723
0.5417	1.9367	180	0.5395
0.5172	2.1507	200	0.5202
0.5033	2.3661	220	0.5089
0.4996	2.5814	240	0.5054
0.5043	2.7968	260	0.5032