It uses the interface of the SageMaker Inference Toolkit as is, so it can be easily deployed to SageMaker Endpoint.
Open-weight model · Text classification
rubert-tiny-toxicity
by David Dale cointegrated/rubert-tiny-toxicity
rubert-tiny-toxicity is an open-weight model for text classification from David Dale, released under MIT License. It has 12M parameters and a 512-token context. At 16-bit it needs about 0 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index. It draws 451.6k downloads a month.
This is the cointegrated/rubert-tiny model fine-tuned for classification of toxicity and inappropriateness for short informal Russian texts, such as comments in social networks.
Parameters12M
Context512
Weights94.3 MB
Licensemit
AccessOpen weights
Monthly Downloads451.6k
Runs On
What it takes to serve rubert-tiny-toxicity (12M 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 | 0.0 GB | 0.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.0 GB | 0.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.0 GB | 0.0 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 Oct 14, 2026.
rubert-tiny-toxicity on every accelerator the SAVRN Index prices, at every precision
SAVRN's Notes on rubert-tiny-toxicity
Moderating Russian-language comments is the job, and the model is small enough to live inside the comment pipeline itself. David Dale trained it to return five labels on a short informal text: non-toxic, insult, obscenity, threat, and dangerous. Twelve million parameters in three layers come to 94 MB of float32 weights, and our page rounds the 16-bit memory need to 0.0 GB. The cheapest listed setup, one MI300X at $1.85 an hour, is more card than this needs; it belongs on a spare slice of something already running.
MIT terms permit commercial use, modification and redistribution provided the copyright and permission notices ride along, so a moderation vendor can ship it inside a product. Check the lineage, it is derived from cointegrated/rubert-tiny, and the 512-token context, which means long threads get split before scoring. It was last updated August 3, 2026, more than four years after its March 2022 release.
Model Card
By David Dale, published under mit, revision 5d37eff84486.
This is the cointegrated/rubert-tiny model fine-tuned for classification of toxicity and inappropriateness for short informal Russian texts, such as comments in social networks.
The problem is formulated as multilabel classification with the following classes:
- non-toxic: the text does NOT contain insults, obscenities, and threats, in the sense of the OK ML Cup competition.
- insult
- obscenity
- threat
- dangerous: the text is inappropriate, in the sense of Babakov et.al., i.e. it can harm the reputation of the speaker.
A text can be considered safe if it is BOTH non-toxic and NOT dangerous.
Usage
The function below estimates the probability that the text is either toxic OR dangerous:
Configuration
- Architecture
- BertForSequenceClassification
- Context length (tokens)
- 512
- Layers
- 3
- Hidden size
- 312
- Feed-forward size
- 600
- Attention heads
- 12
- Vocabulary size
- 29,564
- Stored precision
- float32
- Model type
- bert
Identity and Version
- Repository
- cointegrated/rubert-tiny-toxicity
- Publisher
- David Dale
- Task
- Text classification
- Modality
- Text
- Library
- transformers
- Parameters
- 12M parameters
- Languages
- ru
- Revision
- 5d37eff844868e243467e4c38898bad46c271af2
- First published
- 2022-03-02
- Last updated
- 2026-08-03
Files and Weights
9 files, 95.0 MB in total. The weights are 2 files totalling 94.3 MB in bin, safetensors.
Weights2 files · 94.3 MB
Configuration2 files · 1.1 KB
Tokenizer3 files · 709.5 KB
Documentation1 file · 3.1 KB
Repository1 file · 791 B
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 47.2 MB | e29f120dfe1f |
| pytorch_model.bin | Weights | 47.2 MB | 3acf3c98b1f8 |
| config.json | Configuration | 957 B | — |
| special_tokens_map.json | Configuration | 112 B | — |
| README.md | Documentation | 3.1 KB | — |
| .gitattributes | Repository | 791 B | — |
| tokenizer.json | Tokenizer | 468.0 KB | — |
| tokenizer_config.json | Tokenizer | 377 B | — |
| vocab.txt | Tokenizer | 241.1 KB | — |
License and Download
- License
- mit
- Access
- Open weights, no gate
- Download size
- 94.3 MB
Download from David Dale
Released by David Dale through its official repository on Hugging Face. Read the license.
Built From
- Derived from cointegrated/rubert-tiny
- Described by arXiv:2103.05345
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 94.3 MB |
| 16-bit | 0.0 GB |
| 8-bit | 0.0 GB |
| 4-bit | 0.0 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About rubert-tiny-toxicity
How much GPU memory does rubert-tiny-toxicity need?
About 0 GB at 16-bit and 0 GB at 4-bit: the weights (12M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run rubert-tiny-toxicity 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.
Can I use rubert-tiny-toxicity commercially?
Yes. rubert-tiny-toxicity is released under MIT License. The MIT License is a short permissive license. It permits commercial use, modification and redistribution, provided the copyright notice and permission notice are included.
What is rubert-tiny-toxicity's context length?
512 tokens, from the maximum position embeddings in its published configuration.
Similar Models
This is the model card of a transformers model that has been pushed on the Hub. Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Use the code below to get started with the model. Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).
Nebulixlabs/Quantum-Classifier 0 = safe 1 = unsafe The held-out test split, where available, is evaluated only after fine-tuning.
This is the model card of a transformers model that has been pushed on the Hub. Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Use the code below to get started with the model. Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).
Baseline test accuracy: 84.49%; Fine-tuned test accuracy: 99.21%.
This is my fine-tuned model for Homework 1. The model classifies HC3 answers as: - 0: human - 1: ChatGPT sentence-transformers/all-MiniLM-L6-v2 I used the fixed dataset splits provided in the notebook.