标签：Transformers

微软的 21 课 GenAI 动手课程，用来替代零散博客教程，把从 prompt 到 RAG 的链路一次走通

配套书的 GPT 从零实现代码，用来把 LLM 训练流程拆开看一遍，不是拿来训生产模型

一门用可运行代码和本地 grader 讲清 LLM 内部、检索与 Agent 运行时的课程，不是框架

把 100+ LLM/VLM 的 SFT、DPO、PPO 收敛到一套配置和 CLI 上，省掉每个模型重写训练脚本

把论文公式翻译成可逐行对照的 PyTorch 代码，用来读算法而不是直接上生产

一个文件一个 ViT 变体的参考实现库，用来读论文代码和搭实验骨架，不是拿来直接上线的模型库

CVPR 2026 录用论文与开源代码的索引清单，用来替代在 arXiv 和 GitHub 之间手动搜索

在 Transformers 模型上注入 LoRA 等适配器，只训练千分之一参数并单独保存适配器权重

一本讲大规模训练与推理运维细节的开源手册，顶替的是团队内部口口相传的排障经验和集群验收清单

在 NVIDIA 多卡集群上训练百亿参数以上 Transformer 时的并行底座，替代手写 TP/PP 拼装

自托管的文档问答与智能体平台，顶替 SaaS 版 ChatGPT 企业知识库

把 Hugging Face 模型推理搬进浏览器或 Node，省掉模型服务这一层

线性时间、常数空间推理的 RNN 大模型，训练时按 Transformer 并行化，用来替掉需要 KV cache 的 Transformer 推理栈

用嵌入索引统一语义检索与 LLM 流水线，替代自己拼向量库加编排层

手写 Transformer 全流程的教学代码库，用来替换 transformers/trl 那套黑盒，看清每一步在算什么

飞桨生态里做大模型微调、量化和多硬件推理的一体化套件，替代 HuggingFace 加 vLLM 那套组合

PyTorch 语音任务的全流程工具箱，用一套 YAML 配方覆盖从微调到推理

在数据出库前做 PII 识别与脱敏的 SDK，顶替手写正则加自研 NER 的那一层

在 PyTorch 里用两行代码搭出带预训练骨干的语义分割网络，省掉自己拼 encoder-decoder 的活

按模型分目录的 Transformers 实战 notebook 集，顶替你自己摸索 API 拼装示例的那一步

训练和推理之间那层文本到 ID 的转换，用 Rust 实现，Python 侧只做薄封装

把训练好的模型编译到 Intel CPU/GPU/NPU 上跑推理，顶替的是各框架自带 runtime 和手写算子优化

把 Whisper + NLLB 塞进安卓手机本地跑，替代联网翻译 API 的离线对话翻译应用

一份 YAML 加层流式，让 4 GB 笔记本显卡也能跑 8B 模型的 LoRA 微调

用 Ray 把训练和 vLLM 推理拆成两个集群，在单机多卡上跑 PPO/GRPO 这类在线 RL 后训练

SANA & Sol: Efficient Generative Models and Full-Stack Training & Inference Engines for Image, Video, and World Modeling

Swap GPT for any LLM by changing a single line of code. Xinference lets you run open-source, speech, and multimodal models on cloud, on-prem, or your laptop — all through one unified, production-ready inference API.

Accelerate local LLM inference and finetuning (LLaMA, Mistral, ChatGLM, Qwen, DeepSeek, Mixtral, Gemma, Phi, MiniCPM, Qwen-VL, MiniCPM-V, etc.) on Intel XPU (e.g., local PC with iGPU and NPU, discrete GPU such as Arc, Flex and Max); seamlessly integrate with llama.cpp, Ollama, HuggingFace, LangChain, LlamaIndex, vLLM, DeepSpeed, Axolotl, etc.

[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". An *ultra-simple, user-friendly yet state-of-the-art* codebase for autoregressive image generation!

A text-to-speech (TTS), speech-to-text (STT) and speech-to-speech (STS) library built on Apple's MLX framework, providing efficient speech analysis on Apple Silicon.

BertViz: Visualize Attention in Transformer Models

Leveraging BERT and c-TF-IDF to create easily interpretable topics.

Implementation of RLHF (Reinforcement Learning with Human Feedback) on top of the PaLM architecture. Basically ChatGPT but with PaLM

An implementation of model parallel autoregressive transformers on GPUs, based on the Megatron and DeepSpeed libraries

SD.Next: All-in-one WebUI for AI generative image and video creation, captioning and processing

Open-source offline translation library written in Python

On-device Speech AI for Apple Silicon

Chronos: Pretrained Models for Time Series Forecasting

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

⚡ Pure-Rust WebGPU inference engine — OpenAI-API compatible, GGUF native, runs on any GPU. No Python. No llama.cpp. Single binary.

A concise but complete full-attention transformer with a set of promising experimental features from various papers

Robust recipes to align language models with human and AI preferences

Superduper: End-to-end framework for building custom AI applications and agents.

A machine learning software for extracting information from scholarly documents

The open source research environment for AI researchers to seamlessly train, evaluate, and scale models from local hardware to GPU clusters.

《大模型白盒子构建指南》：一个全手搓的Tiny-Universe

🌟100+ 原创 LLM / RL 原理图📚，《大模型算法》作者巨献！💥（100+ LLM/RL Algorithm Maps ）

LightGlue: Local Feature Matching at Light Speed (ICCV 2023)

Prompt Engineering | Prompt Versioning | Use GPT or other prompt based models to get structured output. Join our discord for Prompt-Engineering, LLMs and other latest research

⚡️SwanLab - an open-source, modern-design AI training tracking and visualization tool. Supports Cloud / Self-hosted use. Integrated with PyTorch / Transformers / verl / LLaMA Factory / ms-swift / Ultralytics / MMEngine / Keras etc.