File size: 2,957 Bytes
cf85772 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 |
---
library_name: transformers
pipeline_tag: text-generation
inference: true
widget:
- text: Hello!
example_title: Hello world
group: Python
---
This model is for debugging. It is randomly initialized using the config from [Qwen/Qwen2-VL-7B-Instruct](https://huggingface.co/Qwen/Qwen2-VL-7B-Instruct) but with smaller size.
Codes:
```python
import os
from typing import Dict
import requests
import torch
import transformers
from PIL import Image
from torchvision import io
from transformers import (AutoConfig, AutoModelForCausalLM, AutoProcessor,
AutoTokenizer, GenerationConfig, pipeline, set_seed)
from transformers.models.qwen2_vl import Qwen2VLForConditionalGeneration
model_id = "Qwen/Qwen2-VL-7B-Instruct"
repo_id = "yujiepan/qwen2-vl-tiny-random"
save_path = f"/tmp/{repo_id}"
config = AutoConfig.from_pretrained(model_id, trust_remote_code=True)
config.hidden_size = 16
config.intermediate_size = 32
config.num_attention_heads = 2
config.num_hidden_layers = 2
config.num_key_value_heads = 1
config.vision_config.embed_dim = 16
config.vision_config.num_heads = 2
config.vision_config.hidden_size = 16
config.vision_config.depth = 2
config.rope_scaling['mrope_section'] = [1, 1, 2] # sum needs to be 4 here
model = Qwen2VLForConditionalGeneration(config=config)
model = model.to(torch.bfloat16).cuda().eval()
model.generation_config = GenerationConfig.from_pretrained(
model_id, trust_remote_code=True,
)
set_seed(42)
with torch.no_grad():
for _, p in sorted(model.named_parameters()):
torch.nn.init.uniform_(p, -0.3, 0.3)
processor = AutoProcessor.from_pretrained(model_id)
model.save_pretrained(save_path)
processor.save_pretrained(save_path)
os.system(f"ls -alh {save_path}")
def try_inference():
url = "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg"
image = Image.open(requests.get(url, stream=True).raw)
conversation = [
{
"role": "user",
"content": [
{
"type": "image",
},
{"type": "text", "text": "Describe this image."},
],
}
]
processor = AutoProcessor.from_pretrained(save_path)
model = Qwen2VLForConditionalGeneration.from_pretrained(
save_path, torch_dtype=torch.bfloat16, device_map='cuda')
text_prompt = processor.apply_chat_template(
conversation, add_generation_prompt=True)
inputs = processor(
text=[text_prompt], images=[image], padding=True, return_tensors="pt"
)
inputs = inputs.to("cuda")
output_ids = model.generate(**inputs, max_new_tokens=16)
generated_ids = [
output_ids[len(input_ids):]
for input_ids, output_ids in zip(inputs.input_ids, output_ids)
]
output_text = processor.batch_decode(
generated_ids, skip_special_tokens=True, clean_up_tokenization_spaces=True
)
print(output_text)
try_inference()
```
|