PaddlePaddle/PaddleNLP
Easy-to-use and powerful LLM and SLM library with awesome model zoo.
Supports distributed training across 4D parallelism (data, tensor, pipeline, and expert parallel), lossless quantization (FP8, INT8, 4-bit), and high-performance inference with FlashMask operator fusion and dynamic batching. Unified Checkpoint enables flexible resource scaling and 78.5% storage compression, while integrating with PaddlePaddle's framework-specific optimizations like column-sparse attention masking for efficient large model deployment across NVIDIA GPUs, Kunlun XPU, Ascend NPU, and other hardware accelerators.
12,929 stars and 41,348 monthly downloads. Used by 3 other packages. Available on PyPI.
Stars
12,929
Forks
3,060
Language
Python
License
Apache-2.0
Category
Last pushed
Dec 17, 2025
Monthly downloads
41,348
Commits (30d)
0
Dependencies
25
Reverse dependents
3
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/transformers/PaddlePaddle/PaddleNLP"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Compare
Related models
meta-llama/llama-cookbook
Welcome to the Llama Cookbook! This is your go to guide for Building with Llama: Getting started...
arcee-ai/mergekit
Tools for merging pretrained large language models.
changyeyu/LLM-RL-Visualized
๐100+ ๅๅ LLM / RL ๅ็ๅพ๐๏ผใๅคงๆจกๅ็ฎๆณใไฝ่ ๅทจ็ฎ๏ผ๐ฅ๏ผ100+ LLM/RL Algorithm Maps ๏ผ
kyegomez/LFM2
A simple and minimal open source implementation of "Introducing LFM2: The Fastest On-Device...
BeastByteAI/scikit-llm
Seamlessly integrate LLMs into scikit-learn.