Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

New Moondream 2B release! ✨ New features: - Long-form captioning - Open vocab tagging - Better counting, object detection, text understanding - Faster HF transformers inference

vikhyatk's profile picture

vik

28,806 subscribers

51,983 Aufrufe • vor 1 Jahr •via X (Twitter)

BildungWissenschaft & Technologie
LIVE

Anya Rossi• Live Now

Private livecam show

12 Kommentare

Profilbild von vik
vikvor 1 Jahr

Release notes Demo

Profilbild von AssemblyAI
AssemblyAIvor 2 Jahren

Announcing: Our most advanced speech-to-text model goes beyond accuracy to capture the real-world complexity of human conversation and deliver reliable, source-of-truth audio data. Explore Universal-2 updates 👇

Profilbild von Shannon Sands
Shannon Sandsvor 1 Jahr

wen bolting on a diffusion model to an output head and generating ghibli

Profilbild von vik
vikvor 1 Jahr

investigating

Profilbild von snow
snowvor 1 Jahr

this is a really awesome release video btw, i love this format, pretty clean. gonna fit all the new features into moondream-zig :)

Profilbild von vik
vikvor 1 Jahr

possible to integrate xnnpack with zig code? they have good quant matmuls

Profilbild von bellicose_bestie
bellicose_bestievor 1 Jahr

MLX???

Profilbild von vik
vikvor 1 Jahr

soon!

Profilbild von Solsticio
Solsticiovor 1 Jahr

How good is this for OCR?

Profilbild von vik
vikvor 1 Jahr

It was a big focus for this release, but we're only 10% of the way through OCR pretraining. I'd say it's decent but expect a ton more improvement coming soon!

Profilbild von Pratyush 🖇️ life/acc
Pratyush 🖇️ life/accvor 1 Jahr

But can it ghiblify images? 😜

Profilbild von vik
vikvor 1 Jahr

watch this space

Ähnliche Videos

DINO-X: A Unified Vision Model for Open-World Object Detection and Understanding TL;DR: DINO-X Pro: sota model with enhanced perception capabilities for various scenarios; DINO-X Edge: model optimized for faster inference speed and better suited for deployment on edge devices
0:45

DINO-X: A Unified Vision Model for Open-World Object Detection and Understanding TL;DR: DINO-X Pro: sota model with enhanced perception capabilities for various scenarios; DINO-X Edge: model optimized for faster inference speed and better suited for deployment on edge devices

Alexandre Morgand

60,783 Aufrufe • vor 1 Jahr

Releasing moondream-zig! It is a fast, implementation of moondream2 inference on the CPU written from-scratch in Zig :) moondream-zig provides 1.5-2x faster inference compared to huggingface on the same device. moondream vik
1:21

Releasing moondream-zig! It is a fast, implementation of moondream2 inference on the CPU written from-scratch in Zig :) moondream-zig provides 1.5-2x faster inference compared to huggingface on the same device. moondream vik

snow

29,245 Aufrufe • vor 1 Jahr

New Release! Blokees Wheels Transformers C01 Roll Out has arrived. New materials, new thrills. Time to form your own racing team! Blokees Wheels | Transformers | C01 Roll Out #Blokees #BlokeesWheels #Transformers #BlokeesTransformers
1:15

New Release! Blokees Wheels Transformers C01 Roll Out has arrived. New materials, new thrills. Time to form your own racing team! Blokees Wheels | Transformers | C01 Roll Out #Blokees #BlokeesWheels #Transformers #BlokeesTransformers

BLOKEES.OFFICIAL

19,236 Aufrufe • vor 11 Monaten

Gaze detection in the upcoming moondream release. (Live demo in thread ⬇️)
0:38

Gaze detection in the upcoming moondream release. (Live demo in thread ⬇️)

vik

142,682 Aufrufe • vor 1 Jahr

Microsoft's new Florence 2 is big for Computer Vision. It's a merge between Text and Vision. With a single prompt you can instruct the model to do CV tasks like captioning, object detection, grounding, and segmentation. The best part, it only uses a single backbone to handle everything. ▸ Excels in zero-shot performance ▸ Unified model for detection, captioning, etc. ▸ FLD-5B dataset: 5B+ annotations, 126M images ▸ New benchmarks (>5.5+) on COCO, ADE20K
0:27

Microsoft's new Florence 2 is big for Computer Vision. It's a merge between Text and Vision. With a single prompt you can instruct the model to do CV tasks like captioning, object detection, grounding, and segmentation. The best part, it only uses a single backbone to handle everything. ▸ Excels in zero-shot performance ▸ Unified model for detection, captioning, etc. ▸ FLD-5B dataset: 5B+ annotations, 126M images ▸ New benchmarks (>5.5+) on COCO, ADE20K

Lior Alexander

186,652 Aufrufe • vor 2 Jahren

This would have been impossible a year ago. Last week we built a new approach to object detection for Sidequest.xyz: generative spatial understanding. Demo & explanation👇
0:48

This would have been impossible a year ago. Last week we built a new approach to object detection for Sidequest.xyz: generative spatial understanding. Demo & explanation👇

conway

40,487 Aufrufe • vor 1 Jahr

Key features include: -Expanded multimodal understanding with video and audio input, alongside text and images -Developer-friendly sizes: 4B and 2B (and many in between!) -Optimized on-device efficiency for 1.5x faster response on mobile compared to Gemma 3 4B
1:35

Key features include: -Expanded multimodal understanding with video and audio input, alongside text and images -Developer-friendly sizes: 4B and 2B (and many in between!) -Optimized on-device efficiency for 1.5x faster response on mobile compared to Gemma 3 4B

Google AI Developers

11,841 Aufrufe • vor 1 Jahr

This is big. SOTA audio reasoning. SOTA video reasoning. SOTA audio captioning. SOTA sound event detection. Better than Gemini. Better than Qwen. TAC: Timestamped Audio Captioning 📑 paper: 🌐 website with more demos:
0:22

This is big. SOTA audio reasoning. SOTA video reasoning. SOTA audio captioning. SOTA sound event detection. Better than Gemini. Better than Qwen. TAC: Timestamped Audio Captioning 📑 paper: 🌐 website with more demos:

Justin Salamon

19,507 Aufrufe • vor 8 Monaten

⚡️Did you know YOLOv8 is an anchor-free object detection model, and YOLOv5 is an anchor-based object detection model? We conduct an in-depth comparison between them and carry out inference for detection and segmentation. #yolo #yolov8 #yolov5 #ai
0:23

⚡️Did you know YOLOv8 is an anchor-free object detection model, and YOLOv5 is an anchor-based object detection model? We conduct an in-depth comparison between them and carry out inference for detection and segmentation. #yolo #yolov8 #yolov5 #ai

Satya Mallick

84,454 Aufrufe • vor 3 Jahren

✨Cubism Editor 5.0 alpha1 Pre-release✨ Please check the attached digest video for the contents of the major new features. ■New features ■Download and Send feedback #Live2D #cubism5_alpha
1:28

✨Cubism Editor 5.0 alpha1 Pre-release✨ Please check the attached digest video for the contents of the major new features. ■New features ■Download and Send feedback #Live2D #cubism5_alpha

Live2D

124,401 Aufrufe • vor 3 Jahren

✨ Releasing Neodrag v2.0 ✨ - Smaller bundle size! (Up to 15%) - 5 new features - Multiple bug fixes - Faster runtime performance And, the most important: NEW DOCS!
1:33

✨ Releasing Neodrag v2.0 ✨ - Smaller bundle size! (Up to 15%) - 5 new features - Multiple bug fixes - Faster runtime performance And, the most important: NEW DOCS!

{PuruVijay}.svelte

94,732 Aufrufe • vor 3 Jahren

Open source like DeepSeek doesn’t shrink markets. It grows them. Faster, cheaper training means rising inference demand and a new era of AI hardware.
0:47

Open source like DeepSeek doesn’t shrink markets. It grows them. Faster, cheaper training means rising inference demand and a new era of AI hardware.

The Linux Foundation

32,930 Aufrufe • vor 1 Jahr

Excited to release a preview of Moondream 3. A 9B param, 2B active MoE vision language model that makes no compromises; offering state-of-the-art visual reasoning while still retaining an efficient and deployment-friendly form factor.
0:48

Excited to release a preview of Moondream 3. A 9B param, 2B active MoE vision language model that makes no compromises; offering state-of-the-art visual reasoning while still retaining an efficient and deployment-friendly form factor.

vik

228,871 Aufrufe • vor 1 Jahr

BREAKING: The future of object detection is here! Meet YOLO-NAS, open-sourced YOLO-based architecture redefining state-of-the-art in object detection.
1:07

BREAKING: The future of object detection is here! Meet YOLO-NAS, open-sourced YOLO-based architecture redefining state-of-the-art in object detection.

Shubham Saboo

726,668 Aufrufe • vor 3 Jahren

Florence-2, the new vision foundation model by Microsoft, can now run 100% locally in your browser on WebGPU, thanks to Transformers.js! 🤗🤯 It supports tasks like image captioning, optical character recognition, object detection, and many more! 😍 WOW! Demo (+ source code) 👇
1:05

Florence-2, the new vision foundation model by Microsoft, can now run 100% locally in your browser on WebGPU, thanks to Transformers.js! 🤗🤯 It supports tasks like image captioning, optical character recognition, object detection, and many more! 😍 WOW! Demo (+ source code) 👇

Xenova

88,782 Aufrufe • vor 2 Jahren

simultaneous live detection of: 😡 emotion 📕 object 🖐️ finger count single call for 3 detections, 0.2 sec inference each
0:30

simultaneous live detection of: 😡 emotion 📕 object 🖐️ finger count single call for 3 detections, 0.2 sec inference each

Yohei

15,279 Aufrufe • vor 1 Monat

New tiny VLM: LFM2.5-VL-450M > Supports bounding box prediction, object detection, and function calling > Improved multilingual capabilities across 9 languages > Enhanced instruction following for vision and text tasks
1:14

New tiny VLM: LFM2.5-VL-450M > Supports bounding box prediction, object detection, and function calling > Improved multilingual capabilities across 9 languages > Enhanced instruction following for vision and text tasks

Maxime Labonne

32,242 Aufrufe • vor 6 Monaten