Skip to content
Catégories de projets

Apprentissage automatique · GitHub

Explorez les outils pour entraîner, évaluer et exécuter des modèles, et comparez leur popularité.

Projets GitHub utiles

12,832 projets

micronet

666DZY666

2,262étoiles
micronet preview

micronet est une bibliothèque de compression et de déploiement de modèles. La compression comprend : 1) la quantification : entraînement tenant compte de la quantification (QAT), quantification High-Bit (>2b) (DoReFa/Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference), quantification Low-Bit (≤2b) ainsi que quantification ternaire et binaire (TWN/BNN/XNOR-Net), et quantification post-entraînement (PTQ) sur 8 bits (tensorrt) ; 2) l’élagage des canaux des convolutions normales, régulières et groupées ; 3) les structures de convolutions groupées ; et 4) la fusion de la batch normalization pour la quantification. Le déploiement prend en charge TensorRT, fp32/fp16/int8 (ptq-calibration), op-adapt (upsample) et dynamic_shape.

ddddocr

86maid

350étoiles
ddddocr preview

ddddocr rust 版本,ocr_api_server rust 版本,二进制版本,验证码识别,不依赖 opencv 库,跨平台运行,AI MCP 支持,a simple OCR API server, very easy to deploy。

bert_seq2seq

920232796

1,307étoiles
bert_seq2seq preview

pytorch实现 Bert 做seq2seq任务,使用unilm方案,现在也可以做自动摘要,文本分类,情感分析,NER,词性标注等任务,支持t5模型,支持GPT2进行文章续写。

280étoiles
ProteinWorkshop preview

Benchmarking framework for protein representation learning. Includes a large number of pre-training and downstream task datasets, models and training/task utilities. (ICLR 2024)

  • residue_type - One-hot encoding of amino acid type - 21
  • positional_encoding - Transformer-like positional encoding of sequence position - 16

AimYolo

Aa-bN

363étoiles
AimYolo preview

AI外挂——基于YOLOv5的射击类游戏瞄准辅助。An AI plug-in - targeting aid for shooting games based on YOLOv5. (AI自瞄, FPS类游戏适用, 附CSGO专用模型, AI cheating, only for learning)

30étoiles
napari-DINOSim preview

This repository provides DINOSim, a method that leverages the DINOv2 foundation model for zero-shot object detection and segmentation in bioimage analysis. DINOSim uses pretrained DINOv2 embeddings to compare patch similarities, allowing it to detect and segment unseen objects in complex datasets with minimal annotations.

245étoiles

Bayes-Newton—A Gaussian process library in JAX, with a unifying view of approximate Bayesian inference as variants of Newton's method.

260étoiles

Wrapper for a PyTorch classifier which allows it to output prediction sets. The sets are theoretically guaranteed to contain the true class with high probability (via conformal prediction).