File size: 4,133 Bytes
851fb8a |
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 96 97 98 99 100 101 102 103 104 105 |
# This file is part of a work licensed under the Apache License, Version 2.0.
# See the LICENSE file in the root of the original repository:
# https://github.com/LLaVA-VL/LLaVA-NeXT?tab=readme-ov-file
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
# ----------------------------- Modification Notice -----------------------------
# This file was originally obtained from:
# https://github.com/LLaVA-VL/LLaVA-NeXT/blob/4e0ee2d98576210e5a5d122451318d5ef7551fc1/llava/model/multimodal_encoder/siglip_encoder.py#L538-L620
#
# Modification by Yusuke Kanebako on 2025-07-22:
# - Define the Vision Tower for Qwen2-VL based on the Vision Tower definition of SigLip.
import torch
import torch.utils.checkpoint
from torch import nn
from .qwen2_vl.configuration_qwen2_vl import Qwen2VLVisionConfig, Qwen2VLConfig
from .qwen2_vl.modeling_qwen2_vl import Qwen2VLForConditionalGeneration
from .qwen2_vl.image_processing_qwen2_vl import Qwen2VLImageProcessor
from llava.utils import rank0_print
class Qwen2VLVisionTower(nn.Module):
def __init__(self, vision_tower, vision_tower_cfg, delay_load=False):
super().__init__()
self.is_loaded = False
self.config = Qwen2VLVisionConfig()
self.vision_tower_name = vision_tower
self.image_processor = Qwen2VLImageProcessor()
if not delay_load:
rank0_print(f"Loading vision tower: {vision_tower}")
self.load_model()
elif getattr(vision_tower_cfg, "unfreeze_mm_vision_tower", False):
# TODO: better detector is needed.
rank0_print(f"The checkpoint seems to contain `vision_tower` weights: `unfreeze_mm_vision_tower`: True.")
self.load_model()
elif hasattr(vision_tower_cfg, "mm_tunable_parts") and "mm_vision_tower" in vision_tower_cfg.mm_tunable_parts:
rank0_print(f"The checkpoint seems to contain `vision_tower` weights: `mm_tunable_parts` contains `mm_vision_tower`.")
self.load_model()
else:
self.cfg_only = self.config
def load_model(self, device_map=None):
if self.is_loaded:
rank0_print("{} is already loaded, `load_model` called again, skipping.".format(self.vision_tower_name))
return
self.vision_tower = Qwen2VLForConditionalGeneration.from_pretrained(self.vision_tower_name, device_map=device_map).visual
# del self.vision_tower.merger
self.vision_tower.requires_grad_(False)
self.is_loaded = True
def forward(self, images, image_grid_thw=None):
if type(images) is list:
image_features = []
for image in images:
image_feature = self.vision_tower(image.to(device=self.device, dtype=self.dtype).unsqueeze(0), grid_thw=image_grid_thw)
image_features.append(image_feature)
else:
image_features = self.vision_tower(images.to(device=self.device, dtype=self.dtype), grid_thw=image_grid_thw)
return image_features
@property
def dummy_feature(self):
return torch.zeros(1, self.hidden_size, device=self.device, dtype=self.dtype)
@property
def dtype(self):
for p in self.vision_tower.parameters():
return p.dtype
@property
def device(self):
for p in self.vision_tower.parameters():
return p.device
@property
def hidden_size(self):
return self.config.hidden_size
@property
def num_patches(self):
return (self.config.image_size // self.config.patch_size) ** 2
@property
def num_patches_per_side(self):
return self.config.image_size // self.config.patch_size
# return self.model_config["vision_cfg"]["image_size"] // self.model_config["vision_cfg"]["patch_size"]
@property
def image_size(self):
return self.config.image_size
|