Universal Model Framework

Universal Model Framework is a modular neural-model framework built on top of CPUTorch.

The framework is designed to provide the model layer above the computational engine.

Architecture

Universal Model Framework
|
v
CPUTorch
|
v
Tensor / Compute
CPUTorch is the required computational foundation.
The framework does not require a predefined domain dataset.
Users can create their own models and datasets for:
Agriculture
Medicine
Science
Education
Engineering
Finance
Language
Custom domains
Design Principles
1. CPUTorch is the foundation
Universal Model Framework depends on CPUTorch.
The CPUTorch source code is not copied into this project.
2. Domain independent
The framework does not force a specific dataset or knowledge domain.
3. Modular
Different components can evolve independently:
Model
Layers
Dataset
Tokenizer
Trainer
Optimizer
Loss
Evaluation
Checkpoint
4. Scalable
The architecture is designed to support small models as well as larger models, including models targeting approximately 100 million parameters.
Installation
python -m pip install universal-model
CPUTorch will be installed as a dependency.
Development
python examples/basic_model.py
Project Structure
universal_model/
├── core/
├── layers/
├── models/
├── data/
├── training/
├── losses/
├── evaluation/
└── utils/
Roadmap
V0.1
CPUTorch integration foundation
Module abstraction
Parameter abstraction
Model configuration
Base model
Project packaging
V0.2
Linear layer
Embedding
Activations
Normalization
V0.3
Attention
Transformer blocks
Transformer model
V0.4
Dataset API
DataLoader
Tokenizer interface
V0.5
Trainer
Optimizers
Scheduler
Checkpointing
V0.6
Evaluation
Metrics
Parameter counting
Model inspection
V0.7
Target-parameter architecture generation
V1.0
Universal Model Framework.
Author
Frandika Imam Arifin
License
MIT EOF

**Perhatikan:** `EOF` terakhir harus sendirian di barisnya:

```text

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

universal_model-0.1.2.tar.gz (25.1 kB view details)

Uploaded Aug 27, 2026 Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

The dropdown lists show the available interpreters, ABIs, and platforms. Enable javascript to be able to filter the list of wheel files.

universal_model-0.1.2-py3-none-any.whl (34.5 kB view details)

Uploaded Aug 27, 2026 Python 3

File details

Details for the file universal_model-0.1.2.tar.gz.

File metadata

Download URL: universal_model-0.1.2.tar.gz
Upload date: Aug 27, 2026
Size: 25.1 kB
Tags: Source
Uploaded using Trusted Publishing? Yes
Uploaded via: twine/7.0.0 CPython/3.13.14

File hashes

Hashes for universal_model-0.1.2.tar.gz
Algorithm	Hash digest
SHA256	`9140310293d0ae0b87b26b4b7855721febc6051e24d29b113c4fef7e88935a3a`
MD5	`040d75fb356d5e0d75760f8c8a5b6bec`
BLAKE2b-256	`aac139c1fd04ecadf192780877f3fdfaa50099fa31fdd2d6ede0467f39c8e038`

See more details on using hashes here.

Provenance

The following attestation bundles were made for universal_model-0.1.2.tar.gz:

Publisher: publish.yml on Mangono1/universal_model

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Statement:
- Statement type: https://in-toto.io/Statement/v1
- Predicate type: https://docs.pypi.org/attestations/publish/v1
- Subject name: universal_model-0.1.2.tar.gz
- Subject digest: 9140310293d0ae0b87b26b4b7855721febc6051e24d29b113c4fef7e88935a3a
- Sigstore transparency entry: 2617875422
- Sigstore integration time: Aug 27, 2026
Source repository:
- Permalink: Mangono1/universal_model@ae692d60743a347df28c742eb392bf57f4aca7d5
- Branch / Tag: refs/tags/v0.1.2
- Owner: https://github.com/Mangono1
- Access: public
Publication detail:
- Token Issuer: https://token.actions.githubusercontent.com
- Runner Environment: github-hosted
- Publication workflow: publish.yml@ae692d60743a347df28c742eb392bf57f4aca7d5
- Trigger Event: push