ModelVérifiéExterne

Qwen3 VL 2B Sono

It's a huggingface repository of paper "A Multimodal Instruction Dataset and Benchmark for Ultrasound Understanding".

Sourcehuggingface.co/Ssdaizi/Qwen3-VL-2B-Sono/tree/66c256d6d5528b3de147e08eb544c1f5dfed95ef

TypeText modelModèle de baseQwen3VersionmodelÉditeur@SsdaiziCClasse

Source du modèle

Type: Text model
Taille du fichier: 3,96 GB
Modèle de base: Qwen3
Version: vmodel
Mis à jour: 3 oct. 2026
Éditeur: @Ssdaizi
Téléchargements: 22
Paramètres: 2B
Source: Hugging Face

Description de la source

Qwen3-VL-2B-Sono

It's a huggingface repository of paper "A Multimodal Instruction Dataset and Benchmark for Ultrasound Understanding".

🚀 Quick Start

from transformers import Qwen3VLForConditionalGeneration, AutoProcessor
import torch
model_path = "Ssdaizi/Qwen3-VL-2B-Sono" 

# Load the model
model = Qwen3VLForConditionalGeneration.from_pretrained(
    model_path,
    dtype=torch.bfloat16,
    device_map="auto",
)

processor = AutoProcessor.from_pretrained(model_path)

messages = [
    {
        "role": "user",
        "content": [
            {
                "type": "image",
                "image": "test1.png",
            },
            {
                "type": "text",
                "text": "Is this a benign lesion or a malignant lesion?",
            }
        ],
    }
]

# Preparation for inference
inputs = processor.apply_chat_template(
    messages,
    tokenize=True,
    add_generation_prompt=True,
    return_dict=True,
    return_tensors="pt",
)
inputs = inputs.to(model.device)

# Generate response
generated_ids = model.generate(
    **inputs,
    max_new_tokens=512,
    top_p=0.8,
    top_k=20,
    temperature=0.7,
    repetition_penalty=1.0
)

# Remove input tokens from generated output
generated_ids_trimmed = [
    out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
]

output_text = processor.batch_decode(
    generated_ids_trimmed,
    skip_special_tokens=True,
    clean_up_tokenization_spaces=False,
)

print(output_text[0])

Sources

1 source

Hugging FacePrincipalVersion 66c256d6d5528b3de147e08eb544c1f5dfed95ef

huggingface.co/Ssdaizi/Qwen3-VL-2B-Sono/tree/66c256d6d5528b3de147e08eb544c1f5dfed95ef

Vérifié 3 oct.

Artefacts du modèle

1 artefact

model.safetensors

safetensors · 3,96 GB · SHA-256 d38d7019545e…8d74 · Hugging Face

Download file

Extraits de sources

3 extraits

Readme11 juil.huggingface.co/Ssdaizi/Qwen3-VL-2B-Sono

--- license: apache-2.0 --- # Qwen3-VL-2B-Sono It's a huggingface repository of paper "A Multimodal Instruction Dataset and Benchmark for Ultrasound Understanding". ### 🚀 Quick Start ```python from transformers import Qwen3VLForConditionalGeneration, AutoProcessor import torch model_path = "Ssdaizi/Qwen3-VL-2B-Sono" # Load the model model = Qwen3VLForConditionalGeneration.from_pretrained( model_path, dtype=torch.bfloat16, device_map="auto", ) processor = AutoProcessor.from_pretrained(model_path) messages = [ { "role": "user", "content": [ { "type": "image", "image": "test1.png", }, { "type": "text", "text": "Is this a benign lesion or a malignant lesion?", } ], } ] # Preparation for inference inputs = processor.apply_chat_template( messages, tokenize=True, add_generation_prompt=True, return_dict=True, return_tensors="pt", ) inputs = inputs.to(model.device) # Generate response generated_ids = model.generate( **inputs, max_new_tokens=512, top_p=0.8, top_k=20, temperature=0.7, repetition_penalty=1.0 ) # Remove input tokens from generated output generated_ids_trimmed = [ out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids) ] output_text = processor.batch_decode( generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False, ) print(output_text[0]) ```

Note3 oct.huggingface.co/Ssdaizi/Qwen3-VL-2B-Sono

Source context: 22 downloads · 0 likes · Repo Ssdaizi/Qwen3-VL-2B-Sono

Note17 juil.huggingface.co/Ssdaizi/Qwen3-VL-2B-Sono

Source context: 130 downloads · 0 likes · Repo Ssdaizi/Qwen3-VL-2B-Sono