The Sovereign-Swarm-Coherence-v1-alpha repository contains the lightweight ONNX & WebAssembly (WASM) client-side package for running inference directly inside web browsers and mobile apps with 0ms server latency and zero cloud API infrastructure costs. - 100% Offline Capable: Zero network dependencies after initial WASM module download. Apache 2.0 License © ItsNotAI LABS
Open-weight model · Feature extraction
Sovereign-Swarm-Coherence-v1
by Alfredo Medina ItsnotAilabs/Sovereign-Swarm-Coherence-v1
Sovereign-Swarm-Coherence-v1 is an open-weight model for feature extraction from Alfredo Medina, released under Apache License 2.0. It has 1M parameters. 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 125 downloads a month.
The Sovereign-Swarm-Coherence-v1 is a production-verified PyTorch Multi-Head Self-Attention Transformer model designed for Robotics & Micro-Drone Swarm Control.
Parameters1M
Context—
Weights11.5 MB
Licenseapache-2.0
AccessOpen weights
Monthly Downloads125
Runs On
What it takes to serve Sovereign-Swarm-Coherence-v1 (1M 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 7, 2026.
Sovereign-Swarm-Coherence-v1 on every accelerator the SAVRN Index prices, at every precision
Model Card
By Alfredo Medina, published under apache-2.0, revision 461b1355d60a.
Sovereign Swarm Coherence Transformer v1
Published by ItsNotAI LABS (Dallas, Texas)
The Sovereign-Swarm-Coherence-v1 is a production-verified PyTorch Multi-Head Self-Attention Transformer model designed for Robotics & Micro-Drone Swarm Control.
Mathematical Physics & Explicit Parameter Breakdown
Unlike generic models with arbitrary weight reporting, this repository explicitly itemizes learned trainable parameters versus non-trainable positional encoding constants:
- Trainable Learned Parameters (
requires_grad=True):800,019 - Positional Encoding Constant Buffer Elements (
pos_encoder.pe):640,000 - Total Model State Tensor Elements:
1,440,019 - Checkpoint File Size:
5.51 MB - Trained Optimizer:
AdamW(10 Epochs over domain datasets)
Governing Mathematical Formulation
$$R = \left| \frac{1}{N} \sum_{j=1}^{N} e^{i \theta_j} \right|$$
Primary Use Cases & Capabilities
- Computes exact Kuramoto phase order parameter R and 2D Euclidean pairwise collision risk geometry for micro-drone swarms.
- Domain Application: Real-time multi-agent autonomous swarm synchronization, 0.5m collision proximity alerts, and phase coupling strength adjustment.
- Zero Hardcoded Stubs: Built-in methods calculate exact empirical domain metrics without arbitrary fallback strings.
Configuration
- Architecture
- SovereignSwarmTransformer
- Stored precision
- float32
- Model type
- transformer
Identity and Version
- Repository
- ItsnotAilabs/Sovereign-Swarm-Coherence-v1
- Publisher
- Alfredo Medina
- Task
- Feature extraction
- Modality
- Text
- Library
- Not stated by the source
- Parameters
- 1M parameters
- Languages
- Not stated by the source
- Revision
- 461b1355d60abf1abcdad49962ef8014536647ad
- First published
- 2026-09-11
- Last updated
- 2026-09-20
Files and Weights
11 files, 17.9 MB in total. The weights are 2 files totalling 11.5 MB in bin, safetensors.
Weights2 files · 11.5 MB
Configuration4 files · 514.9 KB
Documentation1 file · 2.3 KB
Other3 files · 5.8 MB
Repository1 file · 1.6 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 5.8 MB | 59778292ce48 |
| pytorch_model.bin | Weights | 5.8 MB | 6737bba7ecf4 |
| agent_helper.py | Configuration | 5.5 KB | — |
| config.json | Configuration | 280 B | — |
| metrics.json | Configuration | 519 B | — |
| weights_dump.json | Configuration | 508.6 KB | — |
| README.md | Documentation | 2.3 KB | — |
| __pycache__/agent_helper.cpython-311.pyc | Other | 11.4 KB | — |
| domain_knowledge_base.sqlite | Other | 12.3 KB | — |
| model.onnx.data | Other | 5.8 MB | 1bdf54c759a9 |
| .gitattributes | Repository | 1.6 KB | — |
License and Download
- License
- apache-2.0
- Access
- Open weights, no gate
- Download size
- 11.5 MB
Download from Alfredo Medina
Released by Alfredo Medina through its official repository on Hugging Face. Read the license.
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 11.5 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 Sovereign-Swarm-Coherence-v1
How much GPU memory does Sovereign-Swarm-Coherence-v1 need?
About 0 GB at 16-bit and 0 GB at 4-bit: the weights (1M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run Sovereign-Swarm-Coherence-v1 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 Sovereign-Swarm-Coherence-v1 commercially?
Yes. Sovereign-Swarm-Coherence-v1 is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.
Similar Models
https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2 with ONNX weights to be compatible with Transformers.js. If you haven't already, you can install the Transformers.js JavaScript library from NPM using: You can then use the model to compute embeddings like this: You can convert this Tensor to a nested JavaScript array using.tolist(): Note: Having a separate repo for ONNX weights is intended to be a temporary solution until WebML gains more traction. If you would like to make your models web-ready, we recommend converting to ONNX using Optimum and structuring your repo like this one (with ONNX weights located in a subfolder named onnx).
https://huggingface.co/BAAI/bge-base-en-v1.5 with ONNX weights to be compatible with Transformers.js. If you haven't already, you can install the Transformers.js JavaScript library from NPM using: You can then use the model to compute embeddings, as follows: You can also use the model for retrieval. For example: Note: Having a separate repo for ONNX weights is intended to be a temporary solution until WebML gains more traction. If you would like to make your models web-ready, we recommend converting to ONNX using Optimum and structuring your repo like this one (with ONNX weights located in a subfolder named onnx).
The abstract of the paper states that: You can use this model for zero shot audio classification or extracting audio and/or textual features. You can also get the audio and text embeddings using ClapModel If you are using this model for your work, please consider citing the original paper
For more details please refer to our Github: FlagEmbedding. If you are looking for a model that supports more languages, longer texts, and other retrieval methods, you can try using bge-m3. FlagEmbedding focuses on retrieval-augmented LLMs, consisting of the following projects currently: - 1/30/2024: Release BGE-M3, a new member to BGE model series! M3 stands for Multi-linguality (100+ languages), Multi-granularities (input length up to 8192), Multi-Functionality (unification of dense, lexical, multi-vec/colbert retrieval). It is the first embedding model which supports all three retrieval methods, achieving new SOTA on multi-lingual (MIRACL) and cross-lingual (MKQA) benchmarks. Technical…
The large model pretrained on 16kHz sampled speech audio. When using the model, make sure that your speech input is also sampled at 16kHz. Note: This model does not have a tokenizer as it was pretrained on audio alone. In order to use this model speech recognition, a tokenizer should be created and the model should be fine-tuned on labeled text data. Check out this blog for more in-detail explanation of how to fine-tune the model. - 60,000 hours of Libri-Light - 10,000 hours of GigaSpeech - 24,000 hours of VoxPopuli Authors: Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin…