Embeddings Vector Formatter
Paste raw numbers or a messy array and instantly generate a formatted JSON payload ready for Pinecone or Milvus. Automatically extracts floats and counts dimensions. 100% client-side.
- Automatic dimension counting
- Extracts numbers from dirty data
- No upload, no sign-up
- Runs in your browser
- No upload
- No sign-up
- Auto dimension count
- Works offline
Related tools
Free AI and developer tools that run in your browser. Nothing uploaded, ever.
Why this tool is the best choice
Stop writing custom Python scripts just to format a vector array for your database.
| Feature | TToolDeck | Python Scripts | Other Web Tools |
|---|---|---|---|
| Runs entirely in your browser | Yes | No | No |
| Vectors never sent to a server | Yes | No | No |
| Automatic dimension counting | Yes | No | No |
| Extracts numbers from messy text | Yes | No | Partial |
| No Python/Node.js required | Yes | No | Yes |
| Free without paywalls | Yes | Yes | Partial |
How it works
- 1
Paste your vector
Copy the raw output from OpenAI, Cohere, or HuggingFace. It can be comma-separated, in brackets, or mixed with text.
- 2
Automatic extraction
The tool finds every number, ignores brackets and text, converts them to floats, and counts the dimensions.
- 3
Copy the JSON
Get a perfectly formatted payload: {"vector": [...], "dimensions": 1536} ready for Pinecone or Milvus.
Everything you get for free
Pinecone & Milvus Ready
Outputs the exact JSON structure required by the most popular vector databases for upserting data.
Auto Dimension Count
You do not need to manually count 1536 numbers. The tool calculates the dimension count automatically and includes it in the output.
100% Secure
Your embeddings represent proprietary data. Because this runs locally, your vectors are never uploaded to a server.
Dirty Data Handling
If your AI output includes text like "Here is the vector: [0.1, 0.2]", we strip the text and extract only the floats.
Scientific Notation
We support scientific notation (e.g., 1.2e-5), which is common in some model outputs.
Zero Dependencies
No need to write a Python script or install pandas. Just paste and copy.
What people use it for
Upserting to Pinecone
Quickly format a vector before making a POST request to the Pinecone API to insert a new document.
Milvus Insertions
Convert raw arrays into the JSON format required for Milvus bulk insert operations.
Debugging Vectors
Quickly see how many dimensions a model actually returned when you suspect an API error.
Preparing Test Data
Format mock vectors to seed your test database without hitting the OpenAI API.
Data Migration
Moving between vector databases? Format the raw vectors into the new database's expected structure.
Educational
Understand the relationship between an embedding model's output length and the required database index dimension.
Questions people ask
What formats does this support?
It accepts raw comma-separated numbers, standard JSON arrays (like [0.1, 0.2]), Python lists, or even text that has numbers mixed with other words. It extracts every number it finds.
Why does it add the "dimensions" field?
Many vector databases (like Pinecone) require you to specify the dimension count when creating an index. Adding it to the JSON payload makes it easier to verify before you run your script.
Does this generate the embeddings themselves?
No. This is a formatting tool. You must generate the embeddings using OpenAI, Cohere, or HuggingFace first. This tool takes that raw output and formats it for database ingestion.
Does it handle floats and scientific notation?
Yes. The regex is designed to capture standard floats, negative numbers, and scientific notation (e.g., 1.23e-4) common in machine learning outputs.
Is there a limit on vector size?
Because it runs in the browser, it can handle standard embedding sizes (up to 3072 dimensions) instantly.
Can I use this for Weaviate?
Yes, the resulting JSON array can be easily adapted for Weaviate's GraphQL or REST API payloads.
Is my data sent to a server?
No. The entire extraction and formatting process happens locally in your browser using JavaScript regex.
What if the text contains years or other numbers?
The tool extracts all numbers it finds. If your raw text contains stray numbers (like dates), you should clean the text or manually remove those numbers before pasting to ensure they are not included in the vector.