Internvl3 8B

🚀 InternVL3-8B

InternVL3-8B is an advanced multimodal large language model that combines vision and language capabilities, offering superior performance in various multimodal and language tasks.

[🐱 GitHub] [📄 InternVL 1.0] [📄 InternVL 1.5] [📄 InternVL 2.5] [📄 InternVL2.5-MPO] [📄 InternVL3]

[📝 Blog] [💬 Chat Demo] [🤗 HF Demo] [🚀 Quick Start] [📚 Documents]

✨ Features

We introduce InternVL3, an advanced multimodal large language model (MLLM) series that demonstrates superior overall performance.

• Enhanced Multimodal Capabilities: Compared to InternVL 2.5, InternVL3 exhibits superior multimodal perception and reasoning capabilities, while further extending its multimodal capabilities to encompass tool usage, GUI agents, industrial image analysis, 3D vision perception, and more.
• Better Language Performance: We compare InternVL3 with Qwen2.5 Chat models, whose corresponding pre-trained base models are employed as the initialization of the language component in InternVL3. Benefitting from Native Multimodal Pre-Training, the InternVL3 series achieves even better overall text performance than the Qwen2.5 series.

📚 Documentation

InternVL3 Family

In the following table, we provide an overview of the InternVL3 series.

Model Architecture

As shown in the following figure, InternVL3 retains the same model architecture as InternVL 2.5 and its predecessors, InternVL 1.5 and 2.0, following the "ViT-MLP-LLM" paradigm. In this new version, we integrate a newly incrementally pre-trained InternViT with various pre-trained LLMs, including InternLM 3 and Qwen 2.5, using a randomly initialized MLP projector.

As in the previous version, we applied a pixel unshuffle operation, reducing the number of visual tokens to one-quarter of the original. Besides, we adopted a similar dynamic resolution strategy as InternVL 1.5, dividing images into tiles of 448×448 pixels. The key difference, starting from InternVL 2.0, is that we additionally introduced support for multi-image and video data.

Notably, in InternVL3, we integrate the Variable Visual Position Encoding (V2PE), which utilizes smaller, more flexible position increments for visual tokens. Benefiting from V2PE, InternVL3 exhibits better long context understanding capabilities compared to its predecessors.

Training Strategy

Native Multimodal Pre-Training

We propose a Native Multimodal Pre-Training approach that consolidates language and vision learning into a single pre-training stage. In contrast to standard paradigms that first train a language-only model and subsequently adapt it to handle additional modalities, our method interleaves multimodal data (e.g., image-text, video-text, or image-text interleaved sequences) with large-scale textual corpora. This unified training scheme allows the model to learn both linguistic and multimodal representations simultaneously, ultimately enhancing its capability to handle vision-language tasks without the need for separate alignment or bridging modules. Please see our paper for more details.

Supervised Fine-Tuning

In this phase, the techniques of random JPEG compression, square loss re-weighting, and multimodal data packing proposed in InternVL2.5 are also employed in the InternVL3 series. The main advancement of the SFT phase in InternVL3 compared to InternVL2.5 lies in the use of higher-quality and more diverse training data. Specifically, we further extend training samples for tool use, 3D scene understanding, GUI operations, long context tasks, video understanding, scientific diagrams, creative writing, and multimodal reasoning.

Mixed Preference Optimization

During Pre-training and SFT, the model is trained to predict the next token conditioned on previous ground-truth tokens. However, during inference, the model predicts each token based on its own prior outputs. This discrepancy between ground-truth tokens and model-predicted tokens introduces a distribution shift, which can impair the model’s Chain-of-Thought (CoT) reasoning capabilities. To mitigate this issue, we employ MPO, which introduces additional supervision from both positive and negative samples to align the model response distribution with the ground-truth distribution, thereby improving reasoning performance. Specifically, the training objective of MPO is a combination of preference loss \(\mathcal{L}{\text{p}}\), quality loss \(\mathcal{L}{\text{q}}\), and generation loss \(\mathcal{L}_{\text{g}}\), which can be formulated as follows:

$$ \mathcal{L}=w_{p}\cdot\mathcal{L}{\text{p}} + w{q}\cdot\mathcal{L}{\text{q}} + w{g}\cdot\mathcal{L}_{\text{g}}, $$

where \(w_{*}\) represents the weight assigned to each loss component. Please see our paper for more details about MPO.

Test-Time Scaling

Test-Time Scaling has been shown to be an effective method to enhance the reasoning abilities of LLMs and MLLMs. In this work, we use the Best-of-N evaluation strategy and employ VisualPRM-8B as the critic model to select the best response for reasoning and mathematics evaluation.

Evaluation on Multimodal Capability

Multimodal Reasoning and Mathematics

OCR, Chart, and Document Understanding

Multi-Image & Real-World Comprehension

Comprehensive Multimodal & Hallucination Evaluation

Visual Grounding

Multimodal Multilingual Understanding

Video Understanding

GUI Grounding

Spatial Reasoning

Evaluation on Language Capability

We compare InternVL3 with Qwen2.5 Chat models, whose corresponding pre-trained base models are employed as the initialization of the language component in InternVL3. Benefitting from Native Multimodal Pre-Training, the InternVL3 series achieves even better overall text performance than the Qwen2.5 series. Please note that the evaluation scores of Qwen2.5 series may differ from those officially reported, as we have adopted the prompt versions provided in the table across all datasets for OpenCompass evaluation.

Ablation Study

Native Multimodal Pre-Training

We conduct experiments on the InternVL2-8B model while keeping its architecture, initialization parameters, and training data entirely unchanged. Traditionally, InternVL2-8B employs a training pipeline that begins with an MLP warmup phase for feature alignment followed by an Instruction Tuning stage. In our experiments, we substitute the conventional MLP warmup phase with a native multimodal pre-training process. This modification isolates the contribution of native multimodal pre-training to the overall multimodal capability of the model.

The evaluation results in the Figure below shows that the model with native multimodal pre-training exhibits performance on most benchmarks that is comparable to the fully multi-stage-trained InternVL2-8B baseline. Furthermore, when followed by instruction tuning on higher-quality data, the model demonstrates further performance gains across evaluated multimodal tasks. These findings underscore the efficiency of native multimodal pre-training in imparting powerful multimodal capabilities to MLLMs.

Mixed Preference Optimization

As shown in the table below, models fine-tuned with MPO demonstrate superior reasoning performance across seven multimodal reasoning benchmarks compared to their counterparts without MPO. Specifically, InternVL3-78B and InternVL3-38B outperform their counterparts by 4.1 and 4.5 points, respectively. Notably, the training data used for MPO is a subset of that used for SFT, indicating that the performance improvements primarily stem from the training algorithm rather than the training data.

Variable Visual Position Encoding

As reported in the table below, the introduction of V2PE leads to significant performance gains across most evaluation metrics. In addition, our ablation studies—by varying the positional increment \( \delta \)—reveal that even for tasks primarily involving conventional contexts, relatively small \( \delta \) values can achieve optimal performance. These findings provide important insights for future efforts aimed at refining position encoding strategies for visual tokens in MLLMs.

📦 Installation

No installation steps provided in the original README.

💻 Usage Examples

Basic Usage

We provide an example code to run InternVL3-8B using transformers.

⚠️ Important Note

Please use transformers>=4.37.2 to ensure the model works normally.

16-bit (bf16 / fp16)

import torch
from transformers import AutoTokenizer, AutoModel
path = "OpenGVLab/InternVL3-8B"
model = AutoModel.from_pretrained(
    path,
    torch_dtype=torch.bfloat16,
    low_cpu_mem_usage=True,
    use_flash_attn=True,
    trust_remote_code=True).eval().cuda()

BNB 8-bit Quantization

import torch
from transformers import AutoTokenizer, AutoModel
path = "OpenGVLab/InternVL3-8B"
model = AutoModel.from_pretrained(
    path,
    torch_dtype=torch.bfloat16,
    load_in_8bit=True,
    low_cpu_mem_usage=True,
    use_flash_attn=True,
    trust_remote_code=True).eval()

Multiple GPUs

The reason for writing the code this way is to avoid errors that occur during multi-GPU inference due to tensors not being on the same device. By ensuring that the first and last layers of the large language model (LLM) are on the same device, we prevent such errors.

import math
import torch
from transformers import AutoTokenizer, AutoModel

def split_model(model_name):
    device_map = {}
    world_size = torch.cuda.device_count()
    config = AutoConfig.from_pretrained(model_path, trust_remote_code=True)
    num_layers = config.llm_config.num_hidden_layers
    # Since the first GPU will be used for ViT, treat it as half a GPU.
    num_layers_per_gpu = math.ceil(num_layers / (world_size - 0.5))
    num_layers_per_gpu = [num_layers_per_gpu] * world_size
    num_layers_per_gpu[0] = math.ceil(num_layers_per_gpu[0] * 0.5)
    layer_cnt = 0
    for i, num_layer in enumerate(num_layers_per_gpu):
        for j in range(num_layer):
            device_map[f'language_model.model.layers.{layer_cnt}'] = i
            layer_cnt += 1
    device_map['vision_model'] = 0
    device_map['mlp1'] = 0
    device_map['language_model.model.tok_embeddings'] = 0
    device_map['language_model.model.embed_tokens'] = 0
    device_map['language_model.output'] = 0
    device_map['language_model.model.norm'] = 0
    device_map['language_model.model.rotary_emb'] = 0
    device_map['language_model.lm_head'] = 0
    device_map[f'language_model.model.layers.{num_layers - 1}'] = 0

    return device_map

path = "OpenGVLab/InternVL3-8B"
device_map = split_model('InternVL3-8B')
model = AutoModel.from_pretrained(
    path,
    torch_dtype=torch.bfloat16,
    low_cpu_mem_usage=True,
    use_flash_attn=True,
    trust_remote_code=True,
    device_map=device_map).eval()

Advanced Usage

import math
import numpy as np
import torch
import torchvision.transforms as T
from decord import VideoReader, cpu
from PIL import Image
from torchvision.transforms.functional import InterpolationMode
from transformers import AutoModel, AutoTokenizer

IMAGENET_MEAN = (0.485, 0.456, 0.406)
IMAGENET_STD = (0.229, 0.224, 0.225)

def build_transform(input_size):
    MEAN, STD = IMAGENET_MEAN, IMAGENET_STD
    transform = T.Compose([
        T.Lambda(lambda img: img.convert('RGB') if img.mode != 'RGB' else img),
        T.Resize((input_size, input_size), interpolation=InterpolationMode.BICUBIC),
        T.ToTensor(),
        T.Normalize(mean=MEAN, std=STD)
    ])
    return transform

def find_closest_aspect_ratio(aspect_ratio, target_ratios, width, height, image_size):
    best_ratio_diff = float('inf')
    best_ratio = (1, 1)
    area = width * height
    for ratio in target_ratios:
        target_aspect_ratio = ratio[0] / ratio[1]
        ratio_diff = abs(aspect_ratio - target_aspect_ratio)
        if ratio_diff < best_ratio_diff:
            best_ratio_diff = ratio_diff
            best_ratio = ratio
        elif ratio_diff == best_ratio_diff:
            if area > 0.5 * image_size * image_size * ratio[0] * ratio[1]:
                best_ratio = ratio
    return best_ratio

def dynamic_preprocess(image, min_num=1, max_num=12, image_size=448, use_thumbnail=False):
    orig_width, orig_height = image.size
    aspect_ratio = orig_width / orig_height

    # calculate the existing image aspect ratio
    target_ratios = set(
        (i, j) for n in range(min_num, max_num + 1) for i in range(1, n + 1) for j in range(1, n + 1) if
        i * j <= max_num and i * j >= min_num)
    target_ratios = sorted(target_ratios, key=lambda x: x[0] * x[1])

    # find the closest aspect ratio to the target
    target_aspect_ratio = find_closest_aspect_ratio(
        aspect_ratio, target_ratios, orig_width, orig_height, image_size)

    # calculate the target width and height
    target_width = image_size * target_aspect_ratio[0]
    target_height = image_size * target_aspect_ratio[1]
    blocks = target_aspect_ratio[0] * target_aspect_ratio[1]

    # resize the image
    resized_img = image.resize((target_width, target_height))
    processed_images = []
    for i in range(blocks):
        box = (
            (i % (target_width // image_size)) * image_size,
            (i // (target_width // image_size)) * image_size,
            ((i % (target_width // image_size)) + 1) * image_size,
            ((i // (target_width // image_size)) + 1) * image_size
        )
        # split the image
        split_img = resized_img.crop(box)
        processed_images.append(split_img)
    assert len(processed_images) == blocks
    if use_thumbnail and len(processed_images) != 1:
        thumbnail_img = image.resize((image_size, image_size))
        processed_images.append(thumbnail_img)
    return processed_images

def load_image(image_file, input_size=448, max_num=12):
    image = Image.open(image_file).convert('RGB')
    transform = build_transform(input_size=input_size)
    images = dynamic_preprocess(image, image_size=input_size, use_thumbnail=True, max_num=max_num)
    pixel_values = [transform(image) for image in images]
    pixel_values = torch.stack(pixel_values)
    return pixel_values

def split_model(model_name):
    device_map = {}
    world_size = torch.cuda.device_count()
    config = AutoConfig.from_pretrained(model_path, trust_remote_code=True)
    num_layers = config.llm_config.num_hidden_layers
    # Since the first GPU will be used for ViT, treat it as half a GPU.
    num_layers_per_gpu = math.ceil(num_layers / (world_size - 0.5))
    num_layers_per_gpu = [num_layers_per_gpu] * world_size
    num_layers_per_gpu[0] = math.ceil(num_layers_per_gpu[0] * 0.5)
    layer_cnt = 0
    for i, num_layer in enumerate(num_layers_per_gpu):
        for j in range(num_layer):

🔧 Technical Details

The technical details are scattered throughout the "Model Architecture" and "Training Strategy" sections, which cover aspects such as the model's architecture design, training methods, and evaluation strategies.

📄 License

• License: other
• License Name: qwen
• License Link: https://huggingface.co/Qwen/Qwen2.5-72B-Instruct/blob/main/LICENSE