{"id":27091561,"url":"https://github.com/mat-kubiak/manual-ml","last_synced_at":"2026-05-06T04:32:19.283Z","repository":{"id":285089984,"uuid":"943478824","full_name":"mat-kubiak/Manual-ML","owner":"mat-kubiak","description":"DIY ML framework (using only NumPy!) + various demos","archived":false,"fork":false,"pushed_at":"2025-03-31T01:16:19.000Z","size":1078,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":1,"default_branch":"main","last_synced_at":"2025-04-06T07:52:27.957Z","etag":null,"topics":["artificial-intelligence","examples","machine-learning","ml","ml-framework","neural-networks","numpy"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/mat-kubiak.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2025-03-05T19:14:43.000Z","updated_at":"2025-03-31T01:16:23.000Z","dependencies_parsed_at":"2025-03-29T13:22:23.102Z","dependency_job_id":"3fd58e68-4e0d-46e7-a234-6887e33eadd6","html_url":"https://github.com/mat-kubiak/Manual-ML","commit_stats":null,"previous_names":["mat-kubiak/manual-ml"],"tags_count":1,"template":false,"template_full_name":null,"purl":"pkg:github/mat-kubiak/Manual-ML","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mat-kubiak%2FManual-ML","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mat-kubiak%2FManual-ML/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mat-kubiak%2FManual-ML/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mat-kubiak%2FManual-ML/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/mat-kubiak","download_url":"https://codeload.github.com/mat-kubiak/Manual-ML/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mat-kubiak%2FManual-ML/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":32678600,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-06T02:33:58.958Z","status":"ssl_error","status_checked_at":"2026-05-06T02:33:39.611Z","response_time":117,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.5:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["artificial-intelligence","examples","machine-learning","ml","ml-framework","neural-networks","numpy"],"created_at":"2025-04-06T07:52:31.657Z","updated_at":"2026-05-06T04:32:19.270Z","avatar_url":"https://github.com/mat-kubiak.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Manual Machine Learning\n\nMNIST written digit recognition (+ various other demos) trained using a DIY ML framework with TensorFlow-inspired API\n\n## Features\n\nCurrently, the framework supports only linear architecture with dense layers.\n\nThe full list of features:\n\n| Type | Features |\n| - | - |\n| Layers | Dense, Activation |\n| Architectures | linear |\n| Optimizers | Random, SGD with momentum, Adam |\n| Losses | MSE, MAE, Categorical Crossentropy |\n| Metrics | MSE, MAE, Accuracy |\n| Activations | Linear, ReLU, LeakyReLU, Sigmoid, Tanh, Sine |\n| Dataset ops. | shuffling, even/uneven batching |\n\n## Roadmap\n\nFeatures that will likely be added in the future: (in order of importance)\n* Dropout layer\n* Conv2D and Flatten layers\n* validation datasets\n* non-linear architecture with Concatenate layer\n* gradient clipping\n* Huber loss\n* AdamW optimizer\n\n## Dependencies\n\nThe framework itself does not rely on anything else than Numpy, Tqdm and standard python packages.\n\nAdditional dependencies are required for launching demos:\n* Pillow\n* Matplotlib\n\n## Launch\n\nTo launch any of the demos, simply clone the repository and execute them:\n\n```sh\ngit clone https://github.com/mat-kubiak/manual-mnist.git\n\ncd manual-mnist\n\n# (browse `demos` dir for more demos)\npython3 -m demos.sin_function\n```\n\n## Demos\n\n### 1. Sine Function Reconstruction\n\nTrains a network to recreate a $sin(10x)$ function from a single coordinate $x \\in [0, 1]$. Demo features a real-time preview of the process.\n\nCode: [`demos/sin_function.py`](demos/sin_function.py)\n\nhttps://github.com/user-attachments/assets/19e969d7-7b87-4dc7-a0bb-6690ba88af27\n\n### 2. Image reconstruction via SIREN\n\nTrains a network to recreate a chosen gray-scale or rgb image of any dimensions from two coordinates x and y: $x,y \\in [-1, 1]$. The network uses periodic sine activations and a special type of weight initialization with accordance to the SIREN approach. (See: [website](https://www.vincentsitzmann.com/siren/), [paper](https://arxiv.org/abs/2006.09661))\n\nCode: [`demos/image_from_coordinates.py`](demos/image_from_coordinates.py) and [`demos/image_from_coordinates_rgb.py`](demos/image_from_coordinates_rgb.py)\n\n__Example gray-scale image (colored with cmap):__ \n\nhttps://github.com/user-attachments/assets/8ef61d7d-5f81-4562-bf19-d10c7a8b75fe\n\n__Example rgb image:__\n\nhttps://github.com/user-attachments/assets/43d97b44-ec67-4dac-832e-0748fb44bb03\n\n## License\n\nThis project is licensed under the terms of the MIT license. See [LICENSE](LICENSE) for more info.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmat-kubiak%2Fmanual-ml","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fmat-kubiak%2Fmanual-ml","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmat-kubiak%2Fmanual-ml/lists"}