You can deploy this model and start earning money today!

dkhursen/
InternVL2-2b-LoRA-300k-drivelm

Name: InternVL2-2b-LoRA-300k-drivelm
Availability: InStock

OpenGVLab/InternVL2-2B

finetune

Views

No views yet

InternVL2-2B — LoRA fine-tuned on DriveLM (300k, official non-i.i.d. split)

Part of the Master's thesis:

Visual Question Answering for Autonomous Driving Dmytro Khursenko, Czech Technical University in Prague, Faculty of Electrical Engineering, 2026. Supervised by Ing. David Hurych, Ph.D. (Valeo) and doc. Georgios Tolias, Ph.D. (CTU FEE).

GitHub · Demo · Pseudo-label dataset

📄 Thesis PDF expected by end of June 2026, following successful defense (CTU FEE).

This is the data-volume baseline: InternVL2-2B fine-tuned with LoRA on the full official DriveLM-nuScenes training split (~300k QA pairs) without distributional realignment.

Key finding: this model uses 12× more training data than LoRA-25k yet scores 6.7 points lower on Final Score (0.493 vs. 0.560), because the official DriveLM split is non-i.i.d. — ~65% of training QA templates are absent from the test set, causing poor generalisation.

This checkpoint contains fully merged weights — the LoRA adapter is already merged into the base model. Load and use it directly with AutoModel.

All released models

Model	HF repo	Final
LoRA-25k	dkhursen/InternVL2-2b-LoRA-25k-drivelm	0.560
LoRA-300k (this model)	dkhursen/InternVL2-2b-LoRA-300k-drivelm	0.493
LoRA-25k + DL-PL 10% ⭐	dkhursen/InternVL2-2b-LoRA-25k_plus_DL-PL-10pct	0.589
LoRA-25k + Oracle annotation	dkhursen/InternVL2-2b-LoRA-25k-drivelm-offline-redcircle-ctag-bkgd	0.775

Pseudo-label dataset: dkhursen/drivelm-pseudo-labels

Results on custom test split

Evaluated on the custom i.i.d. DriveLM-nuScenes test split (3,340 QA pairs).

Metric	Score
Final	0.493
Accuracy	0.339
ChatGPT	0.706
Language	0.412
BLEU-1	0.607
BLEU-2	0.552
BLEU-3	0.501
BLEU-4	0.452
ROUGE-L	0.676
CIDEr	0.323
Match	0.303
Coord	0.006

Score definitions (all table values normalised to [0, 1])

Accuracy — exact-match on MCQ (A/B/C/D) and Yes/No questions; strict letter-only format required

Language — mean of BLEU-1–4, ROUGE-L, and CIDEr/10

ChatGPT — GPT-3.5-turbo semantic similarity judge (0–100 scale, ÷100 in Final)

Match — (F1_coord × 100 + GPT_match) / 2 on prediction answers (0–100 scale, ÷100 in Final); blends spatial coordinate F1 with a GPT judge

Coord — pure coordinate token F1 at L1 < 16 px; diagnostic only, not included in Final

Final — 0.4 × (GPT/100) + 0.2 × Language + 0.2 × (Match/100) + 0.2 × Accuracy

Full metric breakdown: evaluation/README.md

Note on evaluation scope. All scores are measured on the custom i.i.d. local test split (3,340 QA pairs) using the local DriveLM evaluation script. The official DriveLM evaluation server was not used as the primary reporting source: it returns only aggregate scores without per-metric breakdown, the infrastructure was intermittently unreliable (model output parsing errors could not be diagnosed directly — only via GitHub issues), and ChatGPT/Match metrics depend on OpenAI API calls that fail when the API quota for the billing period is exhausted.

Full comparison (all models, online inference):

Model	Final	Acc	ChatGPT	Lang	B1	B2	B3	B4	RL	CIDEr	Match	Coord
Mini-DA†	0.606	0.898	0.668	0.416	0.596	0.564	0.533	0.503	0.651	0.470	0.381	0.000
LoRA-25k	0.560	0.826	0.589	0.459	0.732	0.668	0.606	0.547	0.714	0.230	0.338	0.015
LoRA-25k + DL-PL 10%	0.589	0.836	0.676	0.451	0.719	0.654	0.589	0.525	0.710	0.222	0.304	0.013
LoRA-25k + DL-PL 30%	0.548	0.832	0.605	0.434	0.695	0.625	0.554	0.483	0.692	0.201	0.264	0.008
LoRA-25k + DL-PL 50%	0.532	0.832	0.584	0.433	0.691	0.622	0.551	0.481	0.699	0.171	0.230	0.007
LoRA-25k + DL-PL 100%	0.511	0.805	0.544	0.430	0.687	0.616	0.543	0.470	0.695	0.165	0.232	0.007
LoRA-300k (this model)	0.493	0.339	0.706	0.412	0.607	0.552	0.501	0.452	0.676	0.323	0.303	0.006

† Mini-DA = OpenGVLab/Mini-InternVL2-2B-DA-DriveLM, fine-tuned by OpenGVLab on DriveLM. DL-PL 30%, 50%, 100% variants not publicly released (lower performance without quality filtering).

Training data

Fine-tuned on the official DriveLM-nuScenes training split (~300k QA pairs) as released by OpenDriveLab/DriveLM. No distributional realignment was applied.

The official split is non-i.i.d. due to QA-template distributional mismatch: ~65% of training QA templates are absent from the test set and ~10% of test templates never appear in training, with per-template frequency differences of up to 20 percentage points between splits. This causes models trained on the official split to underperform at test time despite the large data volume. The custom 25k i.i.d. split addresses this; see github.com/dmitrykhursen/VQA-AD-CTU.

Training configuration

Parameter	Value
Base model	OpenGVLab/InternVL2-2B
Fine-tuning method	LoRA (merged into weights in this checkpoint)
LoRA rank	16
LoRA alpha	32
LoRA dropout	0.05
LoRA target modules	all linear layers in the LLM (backbone and MLP frozen)
Learning rate	4e-5 (cosine, warmup 0.03)
Effective batch size	64 (8 GPU × 4 × grad-acc 2)
Epochs	10
Max sequence length	8192
Precision	bfloat16
Hardware	8 × NVIDIA A100 40 GB

Full configuration: configs/finetune/internvl2_2b_lora.yaml

How to use

python

import torch
from PIL import Image
from transformers import AutoTokenizer, AutoModel
import torchvision.transforms as T
from torchvision.transforms.functional import InterpolationMode

MODEL_ID   = "dkhursen/InternVL2-2b-LoRA-300k-drivelm"
IMAGE_PATH = "path/to/stitched_6camera.jpg"
QUESTION   = "<image>\nWhat are the important objects in the current scene? Those objects will be considered for the future reasoning and driving decision."

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


def build_transform(input_size):
    return 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=IMAGENET_MEAN, std=IMAGENET_STD),
    ])


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:
        ratio_diff = abs(aspect_ratio - ratio[0] / ratio[1])
        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
    target_ratios = sorted(
        {(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 min_num <= i * j <= max_num},
        key=lambda x: x[0] * x[1],
    )
    best_ratio = find_closest_aspect_ratio(
        aspect_ratio, target_ratios, orig_width, orig_height, image_size
    )
    target_w = image_size * best_ratio[0]
    target_h = image_size * best_ratio[1]
    cols = best_ratio[0]
    resized = image.resize((target_w, target_h))
    tiles = []
    for i in range(best_ratio[0] * best_ratio[1]):
        col = i % cols
        row = i // cols
        box = (col * image_size, row * image_size,
               (col + 1) * image_size, (row + 1) * image_size)
        tiles.append(resized.crop(box))
    if use_thumbnail and len(tiles) != 1:
        tiles.append(image.resize((image_size, image_size)))
    return tiles


def load_image(image_path, input_size=448, max_num=12):
    image = Image.open(image_path).convert("RGB")
    transform = build_transform(input_size)
    tiles = dynamic_preprocess(image, image_size=input_size, use_thumbnail=True, max_num=max_num)
    return torch.stack([transform(tile) for tile in tiles])


tokenizer = AutoTokenizer.from_pretrained(MODEL_ID, trust_remote_code=True, use_fast=False)
model = AutoModel.from_pretrained(
    MODEL_ID,
    torch_dtype=torch.bfloat16,
    low_cpu_mem_usage=True,
    trust_remote_code=True,
).eval().cuda()

pixel_values = load_image(IMAGE_PATH, max_num=12).to(torch.bfloat16).cuda()
generation_config = {"max_new_tokens": 512, "do_sample": False}

response = model.chat(tokenizer, pixel_values, QUESTION, generation_config)
print("Answer:", response)

Batch inference and evaluation (requires cloning the repo):

bash

git clone https://github.com/dmitrykhursen/VQA-AD-CTU
cd VQA-AD-CTU
bash scripts/setup_env.sh && source vqa-ad-ctu-env/bin/activate

# Run inference (edit MODEL= inside the script)
sbatch scripts/05_inference.sh

# Evaluate
bash scripts/06_evaluate.sh inference/outputs/<MODEL>/local_test.json

Example predictions

Browse 25 sampled scenes: dmitrykhursen.github.io/VQA-AD-CTU

Intended use and limitations

Intended use:

Data-volume vs. distributional alignment ablation from the thesis (Section 4.3)
Reproducing the non-i.i.d. baseline result

Limitations:

Trained on the official non-i.i.d. DriveLM split — generalises poorly to the custom i.i.d. test split
For best performance use dkhursen/InternVL2-2b-LoRA-25k_plus_DL-PL-10pct (Final 0.589)
Not intended for deployment in real autonomous driving systems
Trained and evaluated on nuScenes scenes from Boston and Singapore only

Resources

Resource	Link
Code & scripts	github.com/dmitrykhursen/VQA-AD-CTU
Demo gallery	dmitrykhursen.github.io/VQA-AD-CTU
Pseudo-label dataset	dkhursen/drivelm-pseudo-labels
Best checkpoint	dkhursen/InternVL2-2b-LoRA-25k_plus_DL-PL-10pct
Base model	OpenGVLab/InternVL2-2B
DriveLM benchmark	github.com/OpenDriveLab/DriveLM
nuScenes dataset	nuscenes.org

Citation

bibtex

@mastersthesis{khursenko2026vqa,
  author     = {Khursenko, Dmytro},
  title      = {Visual Question Answering for Autonomous Driving},
  school     = {Czech Technical University in Prague, Faculty of Electrical Engineering},
  year       = {2026},
  supervisor = {Hurych, David and Tolias, Georgios}
}

bibtex

@inproceedings{sima2024drivelm,
  title     = {DriveLM: Driving with Graph Visual Question Answering},
  author    = {Sima, Chonghao and Renz, Katrin and Chitta, Kashyap and others},
  booktitle = {European Conference on Computer Vision (ECCV)},
  year      = {2024}
}