nomic-ai/nomic-embed-text-v1-GGUF

---

Note: For compatiblity with current llama.cpp, please download the files published on 2/15/2024. The files originally published here will fail to load.

---

nomic-embed-text-v1 - GGUF

Original model: nomic-embed-text-v1

Usage

Embedding text with nomic-embed-text requires task instruction prefixes at the beginning of each string.

For example, the code below shows how to use the search_query prefix to embed user questions, e.g. in a RAG application.

To see the full set of task instructions available & how they are designed to be used, visit the model card for nomic-embed-text-v1.5.

Description

This repo contains llama.cpp-compatible files for nomic-embed-text-v1 in GGUF format.

llama.cpp will default to 2048 tokens of context with these files. To use the full 8192 tokens that Nomic Embed is benchmarked on, you will have to choose a context extension method. The original model uses Dynamic NTK-Aware RoPE scaling, but that is not currently available in llama.cpp. A combination of YaRN and linear scaling is an acceptable substitute.

These files were converted and quantized with llama.cpp PR 5500, commit 34aa045de.

Example llama.cpp Command

Compute a single embedding:

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

You can also submit a batch of texts to embed, as long as the total number of tokens does not exceed the context length. Only the first three embeddings are shown by the embedding example.

texts.txt:

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

Compute multiple embeddings:

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

Compatibility

These files are compatible with llama.cpp as of commit 4524290e8 from 2/15/2024.

Provided Files

The below table shows the mean squared error of the embeddings produced by these quantizations of Nomic Embed relative to the Sentence Transformers implementation.

Name	Quant	Size	MSE
nomic-embed-text-v1.Q2_K.gguf	Q2_K	48 MiB	2.36e-03
nomic-embed-text-v1.Q3_K_S.gguf	Q3_K_S	57 MiB	1.31e-03
nomic-embed-text-v1.Q3_K_M.gguf	Q3_K_M	65 MiB	8.73e-04
nomic-embed-text-v1.Q3_K_L.gguf	Q3_K_L	69 MiB	8.68e-04
nomic-embed-text-v1.Q4_0.gguf	Q4_0	75 MiB	6.87e-04
nomic-embed-text-v1.Q4_K_S.gguf	Q4_K_S	75 MiB	6.81e-04
nomic-embed-text-v1.Q4_K_M.gguf	Q4_K_M	81 MiB	3.12e-04
nomic-embed-text-v1.Q5_0.gguf	Q5_0	91 MiB	2.79e-04
nomic-embed-text-v1.Q5_K_S.gguf	Q5_K_S	91 MiB	2.61e-04
nomic-embed-text-v1.Q5_K_M.gguf	Q5_K_M	95 MiB	7.34e-05
nomic-embed-text-v1.Q6_K.gguf	Q6_K	108 MiB	6.29e-05
nomic-embed-text-v1.Q8_0.gguf	Q8_0	140 MiB	6.34e-06
nomic-embed-text-v1.f16.gguf	F16	262 MiB	5.62e-10
nomic-embed-text-v1.f32.gguf	F32	262 MiB	9.34e-11