hjLabs AI Playground
Starting the on-device runtime…
Paraphrase Multilingual MiniLM is a 118M-parameter ONNX model that turns text into a dense vector whose distances encode meaning. On this page it runs entirely inside your browser: the 135 MB download comes straight from the Hugging Face repository “Xenova/paraphrase-multilingual-MiniLM-L12-v2” into your browser cache, and every inference after that executes on your own GPU through WebGPU. That is a one-off download of well under a minute on a normal connection, and small enough that integrated graphics handle it without complaint. The model is multilingual, so it is not restricted to English input.
Puts 50+ languages in one shared vector space, so an English query can match a Hindi document. In practice people reach for a model like this for semantic search, deduplication, clustering and retrieval-augmented generation — all with a local index.
Available precisions: q8 (8-bit) and fp32 (full precision). q8 comes first because it roughly halves the download at a quality cost you will struggle to measure on this class of model; fp32 is there when you want the reference numbers. Hugging Face records around 165k downloads a month — steady, real usage rather than a one-off research drop.
Because there is no inference server behind this page, there is nothing to rate-limit and nothing to bill: the only cost of running Paraphrase Multilingual MiniLM here is the one-off 135 MB download, and once that is cached the page keeps working with the network switched off.
paraphrase multilingual minilmparaphrase multilingual minilm onnxparaphrase multilingual minilm browserrun paraphrase multilingual minilm locallyxenova/paraphrase-multilingual-minilm-l12-v2embeddings modelparaphrase multilingual minilm webgpuparaphrase multilingual minilm download sizefree embeddings modeltransformers.js model
About 135 MB for the q8 build, including the tokenizer and config files the runtime also has to fetch — this is the whole payload, not just the weights file. It is downloaded once and then served from the browser Cache Storage API.
Yes, through the WebAssembly fallback, which is slower but works anywhere. WebGPU (Chrome or Edge 113+, Safari 26+) is what makes it feel interactive. On WASM, prefer the smaller quantized build.
No. Xenova/paraphrase-multilingual-MiniLM-L12-v2 is a public repository, so your browser fetches the ONNX files directly over HTTPS. There is no account, no key and no server in between.
No. The model weights are downloaded from Hugging Face to your browser once, and every inference after that runs on your own GPU through WebGPU. Your text, images and audio are never sent anywhere — hjLabs.in has no inference server and no way to see your input.