造梦西游资源

造梦西游基于pygame的复现demo (单机版/联机版)

LLaMA 2 implemented from scratch in PyTorch

Build qwen2 model from scratch

GPT-SoVITS ONNX Inference Engine & Model Converter

[ICLR'24 spotlight] Chinese and English Multimodal Large Model Series (Chat and Paint) | 基于CPM基础模型的中英双语多模态大模型系列

A CPU Realtime VLM in 500M. Surpassed Moondream2 and SmolVLM. Training from scratch with ease.

A third-party MNN server supporting external calls, embedding model, TTS model and ASR model features.一个支持外部调用、向量模型、文字转语音模型和语音识别模型特性的第三方MNN服务器

整理开源的中文大语言模型，以规模较小、可私有化部署、训练成本较低的模型为主，包括底座模型，垂直领域微调及应用，数据集与教程等。

World's Smallest Vision-Language Model

Everything about the SmolLM and SmolVLM family of models

Official Devkit for the KITTI Depth Prediction/Completion Benchmark 2017

From Big to Small: Multi-Scale Local Planar Guidance for Monocular Depth Estimation

[CVPR 2025 Oral] Reconstruction vs. Generation: Taming Optimization Dilemma in Latent Diffusion Models

Official implementation of "DepthLab: From Partial to Complete"

A family of lightweight multimodal models.

[under review] The official implementation of paper "BrushEdit: All-In-One Image Inpainting and Editing"

LLM大模型（重点）以及搜广推等 AI 算法中手写的面试题，（非 LeetCode），比如 Self-Attention, AUC等，一般比 LeetCode 更考察一个人的综合能力，又更贴近业务和基础知识一点

High-performance Image Tokenizers for VAR and AR

SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformer

Efficient vision foundation models for high-resolution generation and perception.

Janus-Series: Unified Multimodal Understanding and Generation Models

Chinese version of CLIP which achieves Chinese cross-modal retrieval and representation generation.

High-Resolution Image Synthesis with Latent Diffusion Models

[CVPR 2024 - Oral, Best Paper Award Candidate] Marigold: Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation

Autoregressive Model Beats Diffusion: 🦙 Llama for Scalable Image Generation

[NeurIPS 2024 Best Paper Award][GPT beats diffusion🔥] [scaling laws in visual generation📈] Official impl. of "Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction". A…

pytorch ddpm demo

🚀 「大模型」1小时从0训练26M参数的视觉多模态VLM！🌏 Train a 26M-parameter VLM from scratch in just 1 hours!

🚀🚀 「大模型」2小时完全从0训练26M的小参数GPT！🌏 Train a 26M-parameter GPT from scratch in just 2h!