all-MiniLM-L6-v2

all-MiniLM-L6-v2

Turns sentences and short paragraphs into 384-dimensional vectors for semantic search, clustering and sentence similarity. For developers building English search or retrieval features.

Made by sentence-transformers on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: This sentence-transformers model turns a sentence or short paragraph into a vector of 384 numbers. Texts with similar meaning end up close together, which makes the vectors useful for semantic search and clustering. It has 22.7M parameters and is listed for English.

Who it suits: It suits developers who need to compare the meaning of short texts without running a large model, such as people adding search, grouping or duplicate checks to an app.

Industry | Who uses it | How they use it | Example
Software | Developers adding search | Encode documents and queries, then compare the vectors | A help-center search that matches meaning, not only words
Data | Analysts grouping text | Cluster short texts by their vectors | Group support messages by topic
Content | Teams checking overlap | Score how close two sentences are | Flag near-duplicate sentences

People who need to embed long documents or other languages can skip it, since the card is about short English text.

Where to use it:

• Python with the sentence-transformers library
• Python with Hugging Face Transformers and PyTorch
• Windows, macOS or Linux, where you can download the files with Python

Use case | Who | What they get back
Semantic search | Developers | Vectors to rank documents against a query
Clustering | Analysts | Vectors that group similar texts together
Sentence similarity | Developers | A way to compare two sentences by meaning

How to set it up: You need these first:

• Python
• The sentence-transformers package

1. Install the sentence-transformers package with the command below.
2. Optionally download the model files yourself with the Hugging Face tool. They take about 91 MB in F32.
3. Load the model by its id in Python, as shown in the next section.

pip install -U sentence-transformers

pip install -U huggingface_hub
hf download sentence-transformers/all-MiniLM-L6-v2

How to use it: Load the model, pass it a list of sentences and call encode. You get one embedding per sentence. The card gives this script.

from sentence_transformers import SentenceTransformer
sentences = ["This is an example sentence", "Each sentence is converted"]

model = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2')
embeddings = model.encode(sentences)
print(embeddings)

The card also shows a second way that uses Transformers directly. You run the text through the model and then apply mean pooling over the token embeddings yourself.

Tool | What it does
SentenceTransformer | Loads the model by its id
model.encode | Turns a list of sentences into embeddings

Example: You pass in a list of two short sentences and print the result. You get back one vector of 384 numbers for each sentence, ready to compare, cluster or store for search.

For newcomers and experts: An embedding is a list of numbers that stands for the meaning of a text. Two texts about the same thing get lists that sit close together. A good first step is the short script above, which prints the vectors for two sentences.

The model starts from nreimers/MiniLM-L6-H384-uncased and was fine-tuned on a dataset of 1B sentence pairs with a contrastive objective. Given one sentence of a pair, it learns to pick the true partner out of randomly sampled others, using cosine similarity and cross entropy. Training ran on a TPU v3-8 for 100k steps at batch size 1024, with sequences limited to 128 tokens.

Your sentences go in
The model makes one 384-number vector each
You compare or group the vectors

Good to know:

• Size: 22.7M parameters, about 91 MB in F32 and about 11 MB at 4-bit, as an estimate
• Input length: text longer than 256 word pieces is cut off by default
• Versions: Hugging Face lists 132 quantized and 1,164 fine-tuned versions made by other people

Title: Weight size in MB
F32: 91 MB
4-bit estimate: 11 MB

• License: apache-2.0
• Cost to run: none named, no API key or account
• Data it reaches: it downloads model files from Hugging Face, and the script reads only the sentences you pass in
• Not for: long documents, since text beyond 256 word pieces is cut by default

On Hugging Face, 13 November 2026

• Task: sentence similarity
• Library: sentence-transformers
• Parameters: 22.7M
• Weights: about 91 MB in F32; about 11 MB at 4-bit (estimate)
• Formats: Safetensors
• Downloads in the last 30 days: 174,740,184
• Likes: 6,104
• License: apache-2.0
• Languages: en
• Based on: nreimers/MiniLM-L6-H384-uncased
• Quantized versions: 137
• Fine-tunes: 1,165
• Tags: sentence-transformers, pytorch, rust, onnx, safetensors, openvino, bert, feature-extraction, sentence-similarity, transformers, text-embeddings-inference

Links

• Model: https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2
• Files: https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download sentence-transformers/all-MiniLM-L6-v2

Then follow the usage in its model card. Nothing in this kit runs by itself.