--- license: apache-2.0 datasets: - allenai/Molmo2-VideoPoint - allenai/pixmo-points - allenai/pixmo-cap language: - en base_model: - google/siglip-so400m-patch14-384 - Qwen/Qwen3-4B-Instruct-2507 pipeline_tag: video-text-to-text library_name: transformers tags: - multimodal - olmo - molmo - molmo2 --- Logo for the Molmo2 Project # Molmo2-VideoPoint-4B Molmo2 is a family of open vision-language models developed by the Allen Institute for AI (Ai2) that support image, video and multi-image understanding and grounding. Molmo2 models are trained on publicly available third party datasets as referenced in [our technical report](https://allenai.org/papers/molmo2) and [Molmo2 data](https://huggingface.co/collections/allenai/molmo2-data), a collection of datasets with highly-curated image-text and video-text pairs. It has state-of-the-art performance among multimodal models with a similar size. You can find all models in the Molmo2 family [here](https://huggingface.co/collections/allenai/molmo2). **Learn more** about the Molmo2 family [in our announcement blog post](https://allenai.org/blog/molmo2). Molmo2-VideoPoint-4B is based on [Qwen3-4B-Instruct](https://huggingface.co/Qwen/Qwen3-4B-Instruct-2507) and uses [SigLIP 2](https://huggingface.co/google/siglip-so400m-patch14-384) as vision backbone. **Different from the general checkpoints, Molmo2-VideoPoint-4B is finetuned on the Molmo2-VideoPoint data only, after pre-training on pixmo-cap, pixmo-points and tulu's data. It is meant to be used for video pointing and counting only**. Ai2 is commited to open science. The Molmo2 datasets are available [here](https://huggingface.co/collections/allenai/molmo2-data). All other artifacts used in creating Molmo2 (training code, evaluations, intermediate checkpoints) will be made available at a later date, furthering our commitment to open-source AI development and reproducibility. Quick links: - 📂 [All Models](https://huggingface.co/collections/allenai/molmo2) - 📃 [Paper](https://allenai.org/papers/molmo2) - 🎥 [Blog with Videos](https://allenai.org/blog/molmo2) ## Quick Start ### Setup Conda Environment ``` conda create --name transformers4571 python=3.11 conda activate transformers4571 pip install transformers==4.57.1 pip install torch pillow einops torchvision accelerate decord2 molmo_utils ``` ### Pointing Video QA ``` from transformers import AutoProcessor, AutoModelForImageTextToText import torch from molmo_utils import process_vision_info import re model_id="allenai/Molmo2-VideoPoint-4B" # load the processor processor = AutoProcessor.from_pretrained( model_id, trust_remote_code=True, dtype="auto", device_map="auto" ) # load the model model = AutoModelForImageTextToText.from_pretrained( model_id, trust_remote_code=True, dtype="auto", device_map="auto" ) COORD_REGEX = re.compile(rf"<(?:points|tracks).*? coords=\"([0-9\t:;, .]+)\"/?>") FRAME_REGEX = re.compile(rf"(?:^|\t|:|,|;)([0-9\.]+) ([0-9\. ]+)") POINTS_REGEX = re.compile(r"([0-9]+) ([0-9]{3,4}) ([0-9]{3,4})") def _points_from_num_str(text, image_w, image_h, extract_ids=False): all_points = [] for points in POINTS_REGEX.finditer(text): ix, x, y = points.group(1), points.group(2), points.group(3) # our points format assume coordinates are scaled by 1000 x, y = float(x)/1000*image_w, float(y)/1000*image_h if 0 <= x <= image_w and 0 <= y <= image_h: yield ix, x, y def extract_video_points(text, image_w, image_h, extract_ids=False): """Extract video pointing coordinates as a flattened list of (t, x, y) triplets from model output text.""" all_points = [] for coord in COORD_REGEX.finditer(text): for point_grp in FRAME_REGEX.finditer(coord.group(1)): frame_id = float(point_grp.group(1)) w, h = (image_w, image_h) for idx, x, y in _points_from_num_str(point_grp.group(2), w, h): if extract_ids: all_points.append((frame_id, idx, x, y)) else: all_points.append((frame_id, x, y)) return all_points messages = [ { "role": "user", "content": [ dict(type="text", text="Point to the penguins."), dict(type="video", video="https://storage.googleapis.com/oe-training-public/demo_videos/many_penguins.mp4"), ], } ] # process the video using `molmo_utils.process_vision_info` _, videos, video_kwargs = process_vision_info(messages) videos, video_metadatas = zip(*videos) videos, video_metadatas = list(videos), list(video_metadatas) # apply the chat template to the input messages text = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True) # process the video and text inputs = processor( videos=videos, video_metadata=video_metadatas, text=text, padding=True, return_tensors="pt", **video_kwargs, ) inputs = {k: v.to(model.device) for k, v in inputs.items()} # generate output with torch.inference_mode(): generated_ids = model.generate(**inputs, max_new_tokens=2048) # only get generated tokens; decode them to text generated_tokens = generated_ids[0, inputs['input_ids'].size(1):] generated_text = processor.tokenizer.decode(generated_tokens, skip_special_tokens=True) # decode video pointing outputs points = extract_video_points(generated_text, image_w=video_metadatas[0]["width"], image_h=video_metadatas[0]["height"]) print(points) ``` ## Evaluations We report the accuracy and close accuracy on Molmo2-VideoCountEval here. For details on the evals, refer to our [technical report](https://allenai.org/papers/molmo2). | Model | Accuracy | Close Acc. | |-----------------------------|-----------------------------------------|-----------------------------------------| | GPT-5 | 35.8 | 50.3 | | GPT-5 mini | 29.8 | 49.3 | | Gemini 3 Pro | **37.1** | 53.1 | | Gemini 2.5 Pro | 35.8 | **56.5** | | Gemini 2.5 Flash | 31.9 | 48.2 | | Claude Sonnet 4.5 | 27.2 | 45.1 | | Qwen3-VL-4B | 25.3 | 44.3 | | Qwen3-VL-8B | 29.6 | 47.7 | | Molmo2-4B | 34.3 | 56.1 | | Molmo2-8B | 35.5 | 53.3 | | Molmo2-7B | 33.2 | 50.5 | | **Molmo2-VideoPoint-4B (this model)** | 36.8 | **56.5** | ## License and Use This model is licensed under Apache 2.0. It is intended for research and educational use in accordance with Ai2’s [Responsible Use Guidelines](https://allenai.org/responsible-use). This model is trained on third party datasets that are subject to academic and non-commercial research use only. Please review the sources to determine if this model is appropriate for your use case.