AngelSpec
byTencent
A unified, torch-native training framework for MTP and block-parallel speculative decoding.
Machine Learning Model TrainingML FrameworksOpen Source
About
AngelSpec is developed by the Tencent Hunyuan AI Infra team, which is a torch-native framework for training speculative-decoding draft models, covering both autoregressive MTP drafting and the block-parallel DFlash family. It is the training framework behind the technical report: all drafters in the report — the TTT-trained MTP drafter and the DFly family — are trained and released with it.
Open Source Health
- Stars
- 266
- Forks
- 29
- License
- Not stated
- Last commit
- 2 months ago
Python
Alternatives to AngelSpec
Ncnnby Tencentncnn is a high-performance neural network inference framework optimized for the mobile platform
Rclipby Yurij MikhalevichSemantic photo search for the command line
ChineseNERby jingyuanzhangA neural network model for Chinese named entity recognition
Head Pose Estimationby Yin GuobingRealtime human head pose estimation with ONNXRuntime and OpenCV.
Video Subtitle Extractorby 天涯古巷视频硬字幕提取，生成srt文件。无需申请第三方API，本地实现文本识别。基于深度学习的视频字幕提取框架，包含字幕区域检测、字幕内容提取。A GUI tool for extracting hard-coded subtitle (hardsub) from videos and generating srt files.
Resources & Links
Related Categories
Using AngelSpec?
Track its cost next to the rest of your stack and get a reminder before it renews.
Add to my stackVendor
Tencent
Publisher of Bugly and TDesign
