[ICLR'23 Spotlight🔥] The first successful BERT/MAE-style pretraining on any convolutional network; Pytorch impl. of "Designing BERT for Convolutional Networks: Sparse and Hierarchical Masked Modeling"

Firefly中文LLaMA-2大模型，支持增量预训练Baichuan2、Llama2、Llama、Falcon、Qwen、Baichuan、InternLM、Bloom等大模型

An official implementation for " UniVL: A Unified Video and Language Pre-Training Model for Multimodal Understanding and Generation"

BERT-CCPoem is an BERT-based pre-trained model particularly for Chinese classical poetry

Bert-based models(BERT, MTB, CP) for relation extraction.

MatDGL is a neural network package that allows researchers to train custom models for crystal modeling tasks. It aims to accelerate the research and application of material science.

code for paper "Masked Frequency Modeling for Self-Supervised Visual Pre-Training" (https://arxiv.org/pdf/2206.07706.pdf)

Official code repository for the paper "Pushing the Limits of Pre-training for Time Series Forecasting in the CloudOps Domain"

[CCIR 2023] Self-supervised learning for Sequential Recommender Systems

macrogpt大模型全量预训练(1b3,32层), 多卡deepspeed/单卡adafactor

Parsing Hoyoverse game text corpus from public wikipedia

Understanding "A Lite BERT". An Transformer approach for learning self-supervised Language Models.

Running Large Language Model easily.

🎓Automatically Update CV Papers Daily using Github Actions (Update Every 24th hours)