Qwen

/

Qwen2.5-VL-7B-Instruct-AWQ

like 40

Follow

Qwen 22.3k

I have the same issue too. I get an empty string

Description

I'm trying to launch the Qwen2.5-VL-7B-Instruct-AWQ model with the exact example code from your README, but it returns an empty string. I get some initialization warnings about weights that weren't used and others that were newly initialized, followed by completely empty output.

The model loads without errors, but when I try to generate a response for the demo image, it returns an empty list with a single empty string: [''].

Reproduction Steps

1. Installed all required dependencies
2. Used the exact example code from the README with the demo image
3. Executed the code without modifications

Environment

• Python version: Python 3.10.12
• torch version: 2.5.1+cu121
• transformers version: 4.51.0.dev0
• Operating system: Linux (Ubuntu) 22.4
• GPU: NVIDIA Tesla T4 x2
• CUDA version: 12.6

Code

from transformers import Qwen2_5_VLForConditionalGeneration, AutoTokenizer, AutoProcessor
from 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-7B-Instruct-AWQ", 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-7B-Instruct-AWQ",
#     torch_dtype=torch.bfloat16,
#     attn_implementation="flash_attention_2",
#     device_map="auto",
# )

# 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-7B-Instruct-AWQ", 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 inference
text = 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 output
generated_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)

Code Output

Some weights of the model checkpoint at Qwen/Qwen2.5-VL-7B-Instruct-AWQ were not used when initializing Qwen2_5_VLForConditionalGeneration: ['lm_head.weight']
- This IS expected if you are initializing Qwen2_5_VLForConditionalGeneration from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).
- This IS NOT expected if you are initializing Qwen2_5_VLForConditionalGeneration from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).
Some weights of Qwen2_5_VLForConditionalGeneration were not initialized from the model checkpoint at Qwen/Qwen2.5-VL-7B-Instruct-AWQ and are newly initialized: ['lm_head.qweight', 'lm_head.qzeros', 'lm_head.scales']
You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.
Using a slow image processor as `use_fast` is unset and a slow processor was saved with this model. `use_fast=True` will be the default behavior in v4.52, even if the model was saved with a slow processor. This will result in minor differences in outputs. You'll still be able to use a slow processor with `use_fast=False`.
['']

Do you know how can I resolve it?

This issue is reproducible using the sample code directly from the README without modifications.

model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
"Qwen/Qwen2.5-VL-7B-Instruct-AWQ",
torch_dtype=torch.float16,
device_map="auto",
)

i tried with this and i got output. but it still takes close to 46gb. What about you?