Quick Review: SmoothQuant: Accurate and Efficient Post-Training Quantization for LLMs| 东毅居士

Quick Review: SmoothQuant: Accurate and Efficient Post-Training Quantization for LLMs

作者：XD / 发表： 2023年12月7日 00:45 / 更新： 2023年12月7日 00:57 / 科研学习 / 阅读量：773

SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models

Paper: SmoothQuant on arXiv
Code: SmoothQuant on GitHub
Organization: MIT

Highlight:

Hyper-parameter for Outliers: Implements a novel approach using a specific hyper-parameter to manage outliers effectively during the quantization process.

本文作者：XD 转载请标明出处：http://www.eadst.com/blog/229

本站采用知识共享署名-非商业性使用-相同方式共享 4.0 国际许可协议进行许可。

上一篇
Quick Review: AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration

下一篇
Setting Up v2rayNG with Tencent Cloud Silicon Valley Lighthouse

Category

标签云

TensorRT Bipartite Food Github Gemma Website Jetson JSON ModelScope Anaconda scipy RGB Permission Google Plate HaggingFace Vmess VGG-16 NLTK AI HuggingFace Pytorch InvalidArgumentError Web OpenCV Shortcut CEIR NLP XGBoost logger BeautifulSoup Dataset Ptyhon Data Hotel WebCrawler Hungarian Plotly LeetCode git-lfs XML Nginx Template Transformers Vim QWEN PyCharm Knowledge BTC Random Safetensors torchinfo Password Attention Docker Translation Conda SQLite Heatmap Datetime Sklearn FP64 Zip Image2Text Card 搞笑 SPIE Tracking NameSilo Tiktoken Clash Python SVR Django Pandas Pillow Use GGML PIP Cloudreve Color Bert CAM Michelin GoogLeNet Ubuntu EXCEL Bin Git Qwen2.5 GIT GPTQ LaTeX OpenAI Paper FlashAttention 关于博主腾讯云 Quantize PDF OCR Markdown CSV YOLO Video TensorFlow 阿里云 ChatGPT FP32 Input Paddle LLAMA Interview Math Baidu FastAPI BF16 GPT4 Tensor Disk C++ 域名签证 Hilton VSCode Diagram SQL MD5 llama.cpp Pickle TSV git Distillation CUDA Numpy 报税 Proxy Domain uwsgi CV uWSGI Qwen Review FP16 LoRA tar transformers Land Statistics hf DeepStream Quantization UNIX v2ray COCO LLM Crawler Algorithm Windows Base64 ResNet-50 Excel ONNX Augmentation Bitcoin VPN Streamlit Linux 算法题 Magnet CTC diffusers

站点统计

本站现有博文266篇,共被浏览440750次

本站已经建立2019天!

原 Quick Review: SmoothQuant: Accurate and Efficient Post-Training Quantization for LLMs

作者：XD / 发表： 2023年12月7日 00:45 / 更新： 2023年12月7日 00:57 / 科研学习 / 阅读量：773

SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models

Highlight:

Quick Review: SmoothQuant: Accurate and Efficient Post-Training Quantization for LLMs