nomic-embed-text-v1.5-GGUF:支持长文本嵌入的GGUF格式量化模型，适配llama.cpp

0
代码
代码
Issues
Pull Requests
讨论
Wiki
分析
0

Nomic Embed Text V1.5的GGUF格式量化版本，适用于句子相似度与特征提取任务，支持长文本嵌入，提供多种量化级别，可通过llama.cpp高效运行。【此简介由AI生成】

分支2Tags0
JJared Van Bortelcleanup
0188c9bf创建于 2025年4月29日6次提交
文件最后提交记录最后更新时间
.gitattributes
add GGUF quants 2 年前
README.md
cleanup 1 年前
nomic-embed-text-v1.5.Q2_K.ggufLFS
Compatibility with llama.cpp commit 4524290e8 (#2) - compatiblity with llama.cpp PR 5500 (613d71b94d46d8166744aaf54169c4ded6ab5c3d) - update README (f4146037336cfa102d70ce3f5c4f4992a038e6f7) - update README (044c54740261a8f59971db9f2d39f1d1800f9af2) - update README (67ba71ac19871a1d2e5a2236913d9a1878d270fc) 2 年前
nomic-embed-text-v1.5.Q3_K_L.ggufLFS
Compatibility with llama.cpp commit 4524290e8 (#2) - compatiblity with llama.cpp PR 5500 (613d71b94d46d8166744aaf54169c4ded6ab5c3d) - update README (f4146037336cfa102d70ce3f5c4f4992a038e6f7) - update README (044c54740261a8f59971db9f2d39f1d1800f9af2) - update README (67ba71ac19871a1d2e5a2236913d9a1878d270fc) 2 年前
nomic-embed-text-v1.5.Q3_K_M.ggufLFS
Compatibility with llama.cpp commit 4524290e8 (#2) - compatiblity with llama.cpp PR 5500 (613d71b94d46d8166744aaf54169c4ded6ab5c3d) - update README (f4146037336cfa102d70ce3f5c4f4992a038e6f7) - update README (044c54740261a8f59971db9f2d39f1d1800f9af2) - update README (67ba71ac19871a1d2e5a2236913d9a1878d270fc) 2 年前
nomic-embed-text-v1.5.Q3_K_S.ggufLFS
Compatibility with llama.cpp commit 4524290e8 (#2) - compatiblity with llama.cpp PR 5500 (613d71b94d46d8166744aaf54169c4ded6ab5c3d) - update README (f4146037336cfa102d70ce3f5c4f4992a038e6f7) - update README (044c54740261a8f59971db9f2d39f1d1800f9af2) - update README (67ba71ac19871a1d2e5a2236913d9a1878d270fc) 2 年前
nomic-embed-text-v1.5.Q4_0.ggufLFS
Compatibility with llama.cpp commit 4524290e8 (#2) - compatiblity with llama.cpp PR 5500 (613d71b94d46d8166744aaf54169c4ded6ab5c3d) - update README (f4146037336cfa102d70ce3f5c4f4992a038e6f7) - update README (044c54740261a8f59971db9f2d39f1d1800f9af2) - update README (67ba71ac19871a1d2e5a2236913d9a1878d270fc) 2 年前
nomic-embed-text-v1.5.Q4_K_M.ggufLFS
Compatibility with llama.cpp commit 4524290e8 (#2) - compatiblity with llama.cpp PR 5500 (613d71b94d46d8166744aaf54169c4ded6ab5c3d) - update README (f4146037336cfa102d70ce3f5c4f4992a038e6f7) - update README (044c54740261a8f59971db9f2d39f1d1800f9af2) - update README (67ba71ac19871a1d2e5a2236913d9a1878d270fc) 2 年前
nomic-embed-text-v1.5.Q4_K_S.ggufLFS
Compatibility with llama.cpp commit 4524290e8 (#2) - compatiblity with llama.cpp PR 5500 (613d71b94d46d8166744aaf54169c4ded6ab5c3d) - update README (f4146037336cfa102d70ce3f5c4f4992a038e6f7) - update README (044c54740261a8f59971db9f2d39f1d1800f9af2) - update README (67ba71ac19871a1d2e5a2236913d9a1878d270fc) 2 年前
nomic-embed-text-v1.5.Q5_0.ggufLFS
Compatibility with llama.cpp commit 4524290e8 (#2) - compatiblity with llama.cpp PR 5500 (613d71b94d46d8166744aaf54169c4ded6ab5c3d) - update README (f4146037336cfa102d70ce3f5c4f4992a038e6f7) - update README (044c54740261a8f59971db9f2d39f1d1800f9af2) - update README (67ba71ac19871a1d2e5a2236913d9a1878d270fc) 2 年前
nomic-embed-text-v1.5.Q5_K_M.ggufLFS
Compatibility with llama.cpp commit 4524290e8 (#2) - compatiblity with llama.cpp PR 5500 (613d71b94d46d8166744aaf54169c4ded6ab5c3d) - update README (f4146037336cfa102d70ce3f5c4f4992a038e6f7) - update README (044c54740261a8f59971db9f2d39f1d1800f9af2) - update README (67ba71ac19871a1d2e5a2236913d9a1878d270fc) 2 年前
nomic-embed-text-v1.5.Q5_K_S.ggufLFS
Compatibility with llama.cpp commit 4524290e8 (#2) - compatiblity with llama.cpp PR 5500 (613d71b94d46d8166744aaf54169c4ded6ab5c3d) - update README (f4146037336cfa102d70ce3f5c4f4992a038e6f7) - update README (044c54740261a8f59971db9f2d39f1d1800f9af2) - update README (67ba71ac19871a1d2e5a2236913d9a1878d270fc) 2 年前
nomic-embed-text-v1.5.Q6_K.ggufLFS
Compatibility with llama.cpp commit 4524290e8 (#2) - compatiblity with llama.cpp PR 5500 (613d71b94d46d8166744aaf54169c4ded6ab5c3d) - update README (f4146037336cfa102d70ce3f5c4f4992a038e6f7) - update README (044c54740261a8f59971db9f2d39f1d1800f9af2) - update README (67ba71ac19871a1d2e5a2236913d9a1878d270fc) 2 年前
nomic-embed-text-v1.5.Q8_0.ggufLFS
Compatibility with llama.cpp commit 4524290e8 (#2) - compatiblity with llama.cpp PR 5500 (613d71b94d46d8166744aaf54169c4ded6ab5c3d) - update README (f4146037336cfa102d70ce3f5c4f4992a038e6f7) - update README (044c54740261a8f59971db9f2d39f1d1800f9af2) - update README (67ba71ac19871a1d2e5a2236913d9a1878d270fc) 2 年前
nomic-embed-text-v1.5.f16.ggufLFS
Compatibility with llama.cpp commit 4524290e8 (#2) - compatiblity with llama.cpp PR 5500 (613d71b94d46d8166744aaf54169c4ded6ab5c3d) - update README (f4146037336cfa102d70ce3f5c4f4992a038e6f7) - update README (044c54740261a8f59971db9f2d39f1d1800f9af2) - update README (67ba71ac19871a1d2e5a2236913d9a1878d270fc) 2 年前
nomic-embed-text-v1.5.f32.ggufLFS
Compatibility with llama.cpp commit 4524290e8 (#2) - compatiblity with llama.cpp PR 5500 (613d71b94d46d8166744aaf54169c4ded6ab5c3d) - update README (f4146037336cfa102d70ce3f5c4f4992a038e6f7) - update README (044c54740261a8f59971db9f2d39f1d1800f9af2) - update README (67ba71ac19871a1d2e5a2236913d9a1878d270fc) 2 年前

base_model: nomic-ai/nomic-embed-text-v1.5 inference: false language:

  • en license: apache-2.0 model_creator: Nomic model_name: nomic-embed-text-v1.5 model_type: bert pipeline_tag: sentence-similarity quantized_by: Nomic tags:
  • feature-extraction
  • sentence-similarity

nomic-embed-text-v1.5 - GGUF

原始模型：nomic-embed-text-v1.5

使用方法

使用 nomic-embed-text 进行文本嵌入时，需要在每个字符串开头添加任务指令前缀。

例如，以下代码展示了如何在 RAG 应用中使用 search_query 前缀来嵌入用户问题。

如需查看完整的任务指令集及其设计用途，请访问 nomic-embed-text-v1.5 的模型卡片。

说明

本仓库包含适用于 nomic-embed-text-v1.5 的 llama.cpp 兼容文件，格式为 GGUF。

使用这些文件时，llama.cpp 默认上下文长度为 2048 个词元。若需要完整的 8192 词元上下文长度，需选择上下文扩展方法。🤗 Transformers 模型采用动态 NTK 感知 RoPE 缩放技术，但该功能目前尚未在 llama.cpp 中实现。

llama.cpp 命令示例

计算单个嵌入向量：

./embedding -ngl 99 -m nomic-embed-text-v1.5.f16.gguf -c 8192 -b 8192 --rope-scaling yarn --rope-freq-scale .75 -p 'search_query: What is TSNE?'

您也可以提交一批文本来进行嵌入处理，只要总标记数不超过上下文长度限制。embedding示例仅显示前三个嵌入向量。

texts.txt：

search_query: What is TSNE?
search_query: Who is Laurens Van der Maaten?

计算多个嵌入向量：

./embedding -ngl 99 -m nomic-embed-text-v1.5.f16.gguf -c 8192 -b 8192 --rope-scaling yarn --rope-freq-scale .75 -f texts.txt

兼容性

这些文件与截至 2024 年 2 月 15 日提交的 4524290e8 版本的 llama.cpp 兼容。

提供的文件

下表展示了这些量化版本的 Nomic Embed 生成的嵌入向量相对于 Sentence Transformers 实现的均方误差（MSE）。

名称 量化类型 大小 MSE
nomic-embed-text-v1.5.Q2_K.gguf Q2_K 48 MiB 2.33e-03
nomic-embed-text-v1.5.Q3_K_S.gguf Q3_K_S 57 MiB 1.19e-03
nomic-embed-text-v1.5.Q3_K_M.gguf Q3_K_M 65 MiB 8.26e-04
nomic-embed-text-v1.5.Q3_K_L.gguf Q3_K_L 69 MiB 7.93e-04
nomic-embed-text-v1.5.Q4_0.gguf Q4_0 75 MiB 6.32e-04
nomic-embed-text-v1.5.Q4_K_S.gguf Q4_K_S 75 MiB 6.71e-04
nomic-embed-text-v1.5.Q4_K_M.gguf Q4_K_M 81 MiB 2.42e-04
nomic-embed-text-v1.5.Q5_0.gguf Q5_0 91 MiB 2.35e-04
nomic-embed-text-v1.5.Q5_K_S.gguf Q5_K_S 91 MiB 2.00e-04
nomic-embed-text-v1.5.Q5_K_M.gguf Q5_K_M 95 MiB 6.55e-05
nomic-embed-text-v1.5.Q6_K.gguf Q6_K 108 MiB 5.58e-05
nomic-embed-text-v1.5.Q8_0.gguf Q8_0 140 MiB 5.79e-06
nomic-embed-text-v1.5.f16.gguf F16 262 MiB 4.21e-10
nomic-embed-text-v1.5.f32.gguf F32 262 MiB 6.08e-11

项目介绍

Nomic Embed Text V1.5的GGUF格式量化版本，适用于句子相似度与特征提取任务，支持长文本嵌入，提供多种量化级别，可通过llama.cpp高效运行。【此简介由AI生成】

6提交数
定制我的领域
README
规则集

下载使用量

0

项目总下载次数（含Clone、Pull、 zip 包及 release 下载），每日凌晨更新

目录