dung-vpt-uney
Update Visual-CoT demo - 2025-10-12 22:22:10
b20455d
# Copyright 2023 Haotian Liu
#
# 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
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import os
import warnings
import shutil
from transformers import (
AutoTokenizer,
AutoModelForCausalLM,
AutoConfig,
BitsAndBytesConfig,
)
import torch
from llava.model import *
from llava.constants import (
DEFAULT_IMAGE_PATCH_TOKEN,
DEFAULT_IM_START_TOKEN,
DEFAULT_IM_END_TOKEN,
)
def load_pretrained_model(
model_path,
model_base,
model_name,
load_8bit=False,
load_4bit=False,
device_map="auto",
device="cuda",
):
kwargs = {"device_map": device_map}
if load_8bit:
kwargs["load_in_8bit"] = True
elif load_4bit:
kwargs["load_in_4bit"] = True
kwargs["quantization_config"] = BitsAndBytesConfig(
load_in_4bit=True,
bnb_4bit_compute_dtype=torch.bfloat16,
bnb_4bit_use_double_quant=True,
bnb_4bit_quant_type="nf4",
)
else:
kwargs["torch_dtype"] = torch.bfloat16
# Check if model is LLaVA-based (including VisCoT which is built on LLaVA)
if "llava" in model_name.lower() or "viscot" in model_name.lower():
# Load LLaVA model
if "lora" in model_name.lower() and model_base is not None:
raise NotImplementedError
elif model_base is not None:
raise NotImplementedError
else:
tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=False)
model = LlavaLlamaForCausalLM.from_pretrained(
model_path, low_cpu_mem_usage=True, **kwargs
)
else:
raise NotImplementedError
image_processor = None
if "llava" in model_name.lower() or "viscot" in model_name.lower():
mm_use_im_start_end = getattr(model.config, "mm_use_im_start_end", False)
mm_use_im_patch_token = getattr(model.config, "mm_use_im_patch_token", True)
if mm_use_im_patch_token:
tokenizer.add_tokens([DEFAULT_IMAGE_PATCH_TOKEN], special_tokens=True)
if mm_use_im_start_end:
tokenizer.add_tokens(
[DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN], special_tokens=True
)
model.resize_token_embeddings(len(tokenizer))
vision_tower = model.get_vision_tower()
if not vision_tower.is_loaded:
vision_tower.load_model()
vision_tower.to(device=device, dtype=torch.bfloat16)
if hasattr(vision_tower, "image_processor"):
image_processor = vision_tower.image_processor
else:
image_processor = [
vision_tower.image_processor_0,
vision_tower.image_processor_1,
]
if hasattr(model.config, "max_sequence_length"):
context_len = model.config.max_sequence_length
else:
context_len = 2048
return tokenizer, model, image_processor, context_len