240330-大模型资源-使用教程-部署方式-部分笔记

A. 大模型资源

• Models - Hugging Face
• HF-Mirror - Huggingface 镜像站
• 模型库首页 · 魔搭社区

B. 使用教程

HuggingFace

• HuggingFace 10分钟快速入门（一），利用Transformers，Pipeline探索AI。_哔哩哔哩_bilibili
• HuggingFace快速入门（二），利用AI模型快速搭建APP。_哔哩哔哩_bilibili
• HuggingFace 快速入门（三），HF的Agent快速搭建AI Agent_哔哩哔哩_bilibili
• HuggingFace 快速入门（四），玩转HF上的模型_哔哩哔哩_bilibili

HF-Mirror

• 如何快速下载huggingface模型——全方法总结 - 知乎

huggingface-cli download --resume-download mistralai/Mistral-7B-Instruct-v0.2 --local-dir Mistral-7B-Instruct-v0.2 --local-dir-use-symlinks Falsehuggingface-cli download --resume-download google-t5/t5-small --local-dir t5-small --local-dir-use-symlinks Falsehuggingface-cli download --resume-download openai/whisper-large-v3 --local-dir whisper-large-v3 --local-dir-use-symlinks Falsehuggingface-cli download --resume-download openai/whisper-large-v3 --local-dir whisper-large-v3 --local-dir-use-symlinks Falsehuggingface-cli download --resume-download openai/clip-vit-base-patch32 --local-dir clip-vit-base-patch32 --local-dir-use-symlinks Falsehuggingface-cli download --resume-download openai/clip-vit-large-patch14 --local-dir clip-vit-large-patch14 --local-dir-use-symlinks Falsehuggingface-cli download --resume-download keremberke/yolov8m-table-extraction --local-dir yolov8m-table-extraction --local-dir-use-symlinks Falsehuggingface-cli download --resume-download merve/yolov9 --local-dir yolov9 --local-dir-use-symlinks Falsehuggingface-cli download --resume-download stabilityai/stable-code-instruct-3b --local-dir stable-code-instruct-3b --local-dir-use-symlinks Falsehuggingface-cli download --resume-download stabilityai/stable-code-3b --local-dir stable-code-3b--local-dir-use-symlinks Falsehuggingface-cli download --resume-download defog/sqlcoder-7b-2 --local-dir sqlcoder-7b-2 --local-dir-use-symlinks False

魔搭社区

• ModelScope 魔搭社区初探 - 知乎
• 魔搭v1.3版本 Release Note (20230216) · 文档中心

from modelscope.hub.snapshot_download import snapshot_downloadmodel_dir = snapshot_download('iic/speech_fsmn_vad_zh-cn-16k-common-pytorch', cache_dir='speech_fsmn_vad_zh-cn-16k-common-pytorch')model_dir = snapshot_download('qwen/Qwen1.5-MoE-A2.7B', cache_dir='Qwen1.5-MoE-A2.7B')model_dir = snapshot_download('iic/speech_eres2net_large_sv_zh-cn_cnceleb_16k', cache_dir='speech_eres2net_large_sv_zh-cn_cnceleb_16k')model_dir = snapshot_download('iic/cv_ddsar_face-detection_iclr23-damofd', cache_dir='cv_ddsar_face-detection_iclr23-damofd')model_dir = snapshot_download('iic/Whisper-large-v3', cache_dir='Whisper-large-v3-iic')model_dir = snapshot_download('iic/nlp_bart_text-error-correction_chinese', cache_dir='nlp_bart_text-error-correction_chinese')model_dir = snapshot_download('iic/nlp_bart_text-error-correction_chinese-law', cache_dir='nlp_bart_text-error-correction_chinese-law')model_dir = snapshot_download('iic/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch', cache_dir='speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch')model_dir = snapshot_download('iic/cv_dla34_table-structure-recognition_cycle-centernet', cache_dir='cv_dla34_table-structure-recognition_cycle-centernet')model_dir = snapshot_download('iic/cv_resnet-transformer_table-structure-recognition_lore', cache_dir='cv_resnet-transformer_table-structure-recognition_lore') model_dir = snapshot_download('iic/cv_convnextTiny_ocr-recognition-general_damo', cache_dir='cv_convnextTiny_ocr-recognition-general_damo')model_dir = snapshot_download('iic/cv_resnet18_ocr-detection-db-line-level_damo', cache_dir='cv_resnet18_ocr-detection-db-line-level_damo')model_dir = snapshot_download('iic/cv_convnextTiny_ocr-recognition-document_damo', cache_dir='cv_convnextTiny_ocr-recognition-document_damo')model_dir = snapshot_download('iic/cv_resnet18_ocr-detection-line-level_damo', cache_dir='cv_resnet18_ocr-detection-line-level_damo')

C. 部署方式

• 一键部署Google开源大模型Gemma，性能远超Mistral、LLama2 | 本地大模型部署，ollama助您轻松完成！_哔哩哔哩_bilibili
• Windows系统本机运行Gemma最简步骤_哔哩哔哩_bilibili
• ollama如何把gemma模型部署到D盘_哔哩哔哩_bilibili
• Ollama如何把Gemma模型部署到D盘_方法二创建目录链接释放C盘空间_哔哩哔哩_bilibili
• Ollama如何使用现有的GGUF文件_哔哩哔哩_bilibili
• Ollama导入GGUF文件_哔哩哔哩_bilibili
• Ollama wsarecv: An existing connection was forcibly closed by the remote host._哔哩哔哩_bilibili
• OLLAMA_KEEP_ALIVE潜在问题分析_哔哩哔哩_bilibili
• Ollama 运行 GGUF 模型
• LangChain 与 ollama 携手python 环境演示 Hello World 构建属于自己的大模型应用
• 人工智能大模型系列（七）使用 Ollama 和 CodeGPT 在 VSCode 中创建您自己的自定义 Copilot

D. 格式转换

• 开源大模型GGUF量化(llama.cpp)与本地部署运行(ollama)教程 - 知乎

• 揭开gguf神秘面纱——大模型CPU部署系列02 - 知乎

E. 代码生成

• LLM 系列 | 21 : Code Llama实战(上篇) : 模型简介与试用
• LLM系列 | 22 : Code Llama实战(下篇)：本地部署、量化及GPT-4对比
• Stability AI开源3B代码生成模型：可补全，还能Debug

F. YOLOV9

• WongKinYiu/yolov9
• YOLO v9训练自己数据集_yolov9训练自己的数据集-CSDN博客
• 手把手教你在自己的数据集上训练YOLOv9！ - 知乎

G. SAM

• facebookresearch/segment-anything

H. Qwen1.5-MoE

• Github官方GitHub - QwenLM/Qwen1.5: Qwen1.5 is the improved version of Qwen, the large language model series developed by Qwen team, Alibaba Cloud.
• GPTQ版本运行失败Qwen1.5-MoE: 1/3的激活参数量达到7B模型的性能_哔哩哔哩_bilibili
• 对应Chat版本的代码Qwen1.5-MoE: 1/3的激活参数量达到7B模型的性能 - 哔哩哔哩
• 微信资料开源MOE再添一员：通义团队Qwen1.5 MOE A2.7B大模型
• 如何选择模型Qwen1.5系列6个模型如何选择？ AWQ还是GPTQ？#小工蚁_哔哩哔哩_bilibili
• 不同型号显卡在同一台机器上通过vllm加速推理双4090部署qwen72b大模型 每秒150tokens_哔哩哔哩_bilibili
• Qwen-Agent及浏览器插件GitHub - QwenLM/Qwen-Agent: Agent framework and applications built upon Qwen1.5, featuring Function Calling, Code Interpreter, RAG, and Chrome extension.