# Installationpip install "sglang[all]>=0.4.3" --find-links https://flashinfer.ai/whl/cu124/torch2.5/flashinfer-python
# Launchpython3 -m sglang.launch_server --model /Your_Model_Path/DeepSeek-V3-0324 --tp 8 --trust-remote-code

pip uninstall transformerspip install git+https://github.com/huggingface/transformers@3a1ead0aabed473eafe527915eea8c197d424356pip install accelerate

pip install qwen-omni-utils[decord]

import soundfile as sffrom modelscope import Qwen2_5OmniModel, Qwen2_5OmniProcessorfrom qwen_omni_utils import process_mm_info# default: Load the model on the available device(s)# model = Qwen2_5OmniModel.from_pretrained("Qwen/Qwen2.5-Omni-7B", torch_dtype="auto", device_map="auto")# We recommend enabling flash_attention_2 for better acceleration and memory saving.model = Qwen2_5OmniModel.from_pretrained(    "Qwen/Qwen2.5-Omni-7B",    torch_dtype="auto",    device_map="auto",    attn_implementation="flash_attention_2",)processor = Qwen2_5OmniProcessor.from_pretrained("Qwen/Qwen2.5-Omni-7B")conversation = [    {        "role": "system",        "content": "You are Qwen, a virtual human developed by the Qwen Team, Alibaba Group, capable of perceiving auditory and visual inputs, as well as generating text and speech.",    },    {        "role": "user",        "content": [            {"type": "video", "video": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen2.5-Omni/draw.mp4"},        ],    },]# Preparation for inferencetext = processor.apply_chat_template(conversation, add_generation_prompt=True, tokenize=False)audios, images, videos = process_mm_info(conversation, use_audio_in_video=True)inputs = processor(text=text, audios=audios, images=images, videos=videos, return_tensors="pt", padding=True)inputs = inputs.to(model.device).to(model.dtype)# Inference: Generation of the output text and audiotext_ids, audio = model.generate(**inputs, use_audio_in_video=True)text = processor.batch_decode(text_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)print(text)sf.write(    "output.wav",    audio.reshape(-1).detach().cpu().numpy(),    samplerate=24000,)

from modelscope import Qwen2_5_VLForConditionalGeneration, AutoTokenizer, AutoProcessorfrom qwen_vl_utils import process_vision_info
# default: Load the model on the available device(s)model = Qwen2_5_VLForConditionalGeneration.from_pretrained( "Qwen/Qwen2.5-VL-32B-Instruct", torch_dtype="auto", device_map="auto")
# We recommend enabling flash_attention_2 for better acceleration and memory saving, especially in multi-image and video scenarios.# model = Qwen2_5_VLForConditionalGeneration.from_pretrained(# "Qwen/Qwen2.5-VL-32B-Instruct",# torch_dtype=torch.bfloat16,# attn_implementation="flash_attention_2",# device_map="auto",# )
# default processerprocessor = AutoProcessor.from_pretrained("Qwen/Qwen2.5-VL-32B-Instruct")
# The default range for the number of visual tokens per image in the model is 4-16384.# You can set min_pixels and max_pixels according to your needs, such as a token range of 256-1280, to balance performance and cost.# min_pixels = 256*28*28# max_pixels = 1280*28*28# processor = AutoProcessor.from_pretrained("Qwen/Qwen2.5-VL-32B-Instruct", min_pixels=min_pixels, max_pixels=max_pixels)
messages = [ { "role": "user", "content": [ { "type": "image", "image": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg", }, {"type": "text", "text": "Describe this image."}, ], }]
# Preparation for inferencetext = processor.apply_chat_template( messages, tokenize=False, add_generation_prompt=True)image_inputs, video_inputs = process_vision_info(messages)inputs = processor( text=[text], images=image_inputs, videos=video_inputs, padding=True, return_tensors="pt",)inputs = inputs.to("cuda")
# Inference: Generation of the outputgenerated_ids = model.generate(**inputs, max_new_tokens=128)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)