This is a LinkTransformer model. At its core this model this is a sentence transformer model sentence-transformers model- it just wraps around the class. It is designed for quick and easy record linkage (entity-matching) through the LinkTransformer package. The tasks include clustering, deduplication, linking, aggregation and more. Notwithstanding that, it can be used for any sentence similarity task within the sentence-transformers framework as well. It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search. Take a look at the documentation of sentence-transformers if you want to use this model for more than what we…
Open-weight model · Sentence similarity
ko-sroberta-multitask
by Junghyun Gan jhgan/ko-sroberta-multitask
This is a sentence-transformers model: It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search.
Parameters111M
Context514
Weights2.5 GB
License—
AccessOpen weights
Monthly Downloads590.6k
Runs On
What it takes to serve ko-sroberta-multitask (111M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 0.2 GB | 0.3 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.1 GB | 0.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.1 GB | 0.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 18, 2026.
SAVRN's Notes on ko-sroberta-multitask
Korean sentence embeddings are the job here. Junghyun Gan trained this 111M parameter RoBERTa on the KorSTS and KorNLI sets in a multi-task run, and it turns a sentence or paragraph into a 768-dimensional vector for clustering or semantic search. At 16-bit the weights are 0.2 GB and the run needs 0.3 GB, so hardware is not the decision. The cheapest Index slot, one MI300X with 192 GB at $1.85 an hour on demand, leaves nearly the whole card idle; put it beside a larger model on a box you already run.
The license field is blank, and that is the first thing we would settle with the publisher before a paid deployment. Context is 514 tokens, so it encodes passages, not documents. The 2.5 GB download across 21 files is the weights shipped in five formats, safetensors, ONNX, OpenVINO, PyTorch and TF, which opens a CPU-side serving path.
Model Card
This is a sentence-transformers model: It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search. Using this model becomes easy when you have sentence-transformers installed: Then you can use the model like this: Without sentence-transformers, you can use the model like this: First, you pass your input through the transformer model, then you have to apply the right pooling-operation on-top of the contextualized word embeddings. KorSTS, KorNLI 학습 데이터셋으로 멀티 태스크 학습을 진행한 후 KorSTS 평가 데이터셋으로 평가한 결과입니다. The model was trained with the parameters: sentencetransformers.datasets.NoDuplicatesDataLoader.NoDuplicatesDataLoader of…
Excerpt from the card by Junghyun Gan.
Configuration
- Architecture
- RobertaModel
- Context length (tokens)
- 514
- Layers
- 12
- Hidden size
- 768
- Feed-forward size
- 3,072
- Attention heads
- 12
- Vocabulary size
- 32,000
- Stored precision
- float32
- Model type
- roberta
Identity and Version
- Repository
- jhgan/ko-sroberta-multitask
- Publisher
- Junghyun Gan
- Task
- Sentence similarity
- Modality
- Text
- Library
- sentence-transformers
- Parameters
- 111M parameters
- Languages
- ko
- Revision
- 8fca7c9c98c26599be0e14b9916b11a756a26f19
- First published
- 2022-03-02
- Last updated
- 2026-06-16
Files and Weights
21 files, 2.5 GB in total. The weights are 7 files totalling 2.5 GB in bin, h5, onnx, safetensors.
Weights7 files · 2.5 GB
Configuration6 files · 1.5 KB
Tokenizer3 files · 744.1 KB
Documentation1 file · 4.9 KB
Other3 files · 368.9 KB
Repository1 file · 1.2 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 442.5 MB | 5c06458b2668 |
| onnx/model.onnx | Weights | 440.4 MB | 7be16ea0c188 |
| onnx/model_O4.onnx | Weights | 220.2 MB | 3a2c7804d780 |
| onnx/model_qint8_avx512_vnni.onnx | Weights | 111.3 MB | ddd6107a0638 |
| openvino/openvino_model.bin | Weights | 440.1 MB | 008657b34497 |
| pytorch_model.bin | Weights | 442.6 MB | e0143341f5e4 |
| tf_model.h5 | Weights | 442.7 MB | 7a5b1f4865d0 |
| 1_Pooling/config.json | Configuration | 190 B | — |
| config.json | Configuration | 744 B | — |
| config_sentence_transformers.json | Configuration | 123 B | — |
| modules.json | Configuration | 229 B | — |
| sentence_bert_config.json | Configuration | 53 B | — |
| special_tokens_map.json | Configuration | 156 B | — |
| README.md | Documentation | 4.9 KB | — |
| eval/similarity_evaluation_sts-dev_results.csv | Other | 931 B | — |
| openvino/openvino_model.xml | Other | 367.6 KB | — |
| similarity_evaluation_sts-test_results.csv | Other | 302 B | — |
| .gitattributes | Repository | 1.2 KB | — |
| tokenizer.json | Tokenizer | 495.0 KB | — |
| tokenizer_config.json | Tokenizer | 585 B | — |
| vocab.txt | Tokenizer | 248.5 KB | — |
License and Download
- License
- Not stated by the source
- Access
- Open weights, no gate
- Download size
- 2.5 GB
Download from Junghyun Gan
Released by Junghyun Gan through its official repository on Hugging Face.
Built From
- Described by arXiv:2004.03289
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 2.5 GB |
| 16-bit | 0.2 GB |
| 8-bit | 0.1 GB |
| 4-bit | 0.1 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Compare ko-sroberta-multitask
Questions About ko-sroberta-multitask
How much GPU memory does ko-sroberta-multitask need?
About 0.3 GB at 16-bit and 0.1 GB at 4-bit: the weights (111M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run ko-sroberta-multitask on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
What is ko-sroberta-multitask's context length?
514 tokens, from the maximum position embeddings in its published configuration.
Similar Models
This is a LinkTransformer model. At its core this model this is a sentence transformer model sentence-transformers model- it just wraps around the class. It is designed for quick and easy record linkage (entity-matching) through the LinkTransformer package. The tasks include clustering, deduplication, linking, aggregation and more. Notwithstanding that, it can be used for any sentence similarity task within the sentence-transformers framework as well. It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search. Take a look at the documentation of sentence-transformers if you want to use this model for more than what we…
This is a LinkTransformer model. At its core this model this is a sentence transformer model sentence-transformers model- it just wraps around the class. It is designed for quick and easy record linkage (entity-matching) through the LinkTransformer package. The tasks include clustering, deduplication, linking, aggregation and more. Notwithstanding that, it can be used for any sentence similarity task within the sentence-transformers framework as well. It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search. Take a look at the documentation of sentence-transformers if you want to use this model for more than what we…
This is a LinkTransformer model. At its core this model this is a sentence transformer model sentence-transformers model- it just wraps around the class. It is designed for quick and easy record linkage (entity-matching) through the LinkTransformer package. The tasks include clustering, deduplication, linking, aggregation and more. Notwithstanding that, it can be used for any sentence similarity task within the sentence-transformers framework as well. It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search. Take a look at the documentation of sentence-transformers if you want to use this model for more than what we…
This is a LinkTransformer model. At its core this model this is a sentence transformer model sentence-transformers model- it just wraps around the class. It is designed for quick and easy record linkage (entity-matching) through the LinkTransformer package. The tasks include clustering, deduplication, linking, aggregation and more. Notwithstanding that, it can be used for any sentence similarity task within the sentence-transformers framework as well. It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search. Take a look at the documentation of sentence-transformers if you want to use this model for more than what we…
This is a sentence-transformers model: It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search. Using this model becomes easy when you have sentence-transformers installed: Then you can use the model like this: Without sentence-transformers, you can use the model like this: First, you pass your input through the transformer model, then you have to apply the right pooling-operation on-top of the contextualized word embeddings. Text Embeddings Inference (TEI) is a blazing fast inference solution for text embedding models. Send a request to /v1/embeddings to generate embeddings via the OpenAI Embeddings API: Or check…