Feature/improves

.devcontainer/Dockerfile

@@ -1,5 +1,5 @@
 # Use the official .NET 8 SDK image
-FROM mcr.microsoft.com/dotnet/sdk:8.0
+FROM mcr.microsoft.com/dotnet/sdk:10.0
 
 # Set the working directory inside the container
 WORKDIR /workspace

.devcontainer/devcontainer.json

@@ -1,8 +1,11 @@
 {
-    "name": "NET 8 DevContainer",
+    "name": "NET 10 DevContainer",
     "build": {
         "dockerfile": "Dockerfile"
     },
+    "mounts": [
+        "source=${localEnv:USERPROFILE}/.nuget/packages,target=/root/.nuget/packages,type=bind,consistency=cached"
+    ],
     "settings": {
         "terminal.integrated.shell.linux": "/bin/bash"
     },

.github/workflows/ci.yaml

@@ -15,10 +15,10 @@ jobs:
     - name: Checkout
       uses: actions/checkout@main
 
-    - name: Setup .NET 8
+    - name: Setup .NET 10
       uses: actions/setup-dotnet@v1
       with:
-        dotnet-version: '8.0.100'
+        dotnet-version: '10.0.x'
 
     - name: Build
       run: |

.github/workflows/publish.yaml

@@ -15,10 +15,10 @@ jobs:
     - name: Checkout
       uses: actions/checkout@main
 
-    - name: Setup .NET 8
+    - name: Setup .NET 10
       uses: actions/setup-dotnet@v1
       with:
-        dotnet-version: '8.0.100'
+        dotnet-version: '10.0.x'
         source-url: https://nuget.pkg.github.com/${{ github.repository_owner }}/index.json
       env:
         NUGET_AUTH_TOKEN: ${{ github.token }}

.gitignore

@@ -1,8 +1,8 @@
-.idea/
-.vscode/
-
-tiny-brain.sln.DotSettings.user
-
-obj/
-bin/
-
+.idea/
+.vscode/
+
+tiny-brain.sln.DotSettings.user
+
+obj/
+bin/
+

CLAUDE.md

@@ -0,0 +1,63 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Commands
+
+```bash
+# Build entire solution
+dotnet build code/tiny-brain.sln
+
+# Run all tests
+dotnet test code/test/tiny-brain-test.csproj
+
+# Run tests with coverage
+dotnet test code/test/tiny-brain-test.csproj /p:CollectCoverage=true /p:CoverletOutput=TestResults/ /p:CoverletOutputFormat=lcov
+
+# Run a single test by name
+dotnet test code/test/tiny-brain-test.csproj --filter "FullyQualifiedName~<TestName>"
+
+# Run the bigram use-case
+dotnet run --project code/use-cases/biagram/biagram.csproj
+
+# Run the playground
+dotnet run --project code/use-cases/playground/playgroung.csproj
+```
+
+## Architecture
+
+The library is a from-scratch implementation of a neural network with autograd (inspired by micrograd / cs231n). Target: .NET 8, namespace `TinyBrain`.
+
+### Core abstraction: `Operand`
+
+`Operand` (`code/src/Expression/Operand.cs`) is the fundamental building block — a scalar value that carries its gradient and a `_backward` closure. All arithmetic operators (`+`, `-`, `*`, `/`) and unary ops (`Exp`, `Log`, `Pow`, `Relu`, `Tanh`) are overloaded to return new `Operand` instances that record their `Previous` pair. Calling `Backpropagation()` on an `Operand` runs a non-recursive topological sort over the expression graph and then invokes each node's `_backward` in reverse order.
+
+### Neural network layers
+
+```
+Brain  (MLP)
+  └─ Layer[]            (one per hidden/output layer)
+       └─ Neuron[]      (each neuron owns its weights + bias as Operands)
+```
+
+- `Neuron` (`code/src/Neuron/Neuron.cs`) computes `f(Σ(xᵢ·wᵢ) + b)` where `f` is determined by `ActivationType`.
+- `Layer` (`code/src/Layers/Layer.cs`) fans inputs through all neurons; `Forward` returns one `Operand` per neuron.
+- `Brain` (`code/src/MLP/Brain.cs`) chains layers via `Fold`; `Forward` auto-zeros gradients before each pass.
+
+### Activation functions
+
+`Activations.cs` holds the dispatch dictionary keyed on `ActivationType` (currently `None` and `Tanh`). `Tanh` is expressed in terms of `Exp`, `/`, and `-` on `Operand` so its gradient flows automatically through the graph — no hand-coded derivative needed.
+
+### Functional style
+
+The library depends on `tiny-fp` (functional primitives: `Option`, `Unit`, `Map`, `Tee`, `Fold`, `ForEach`). Prefer these combinators over imperative loops; `Unit` is used as the return type of side-effecting void-like operations.
+
+### Use-cases
+
+`code/use-cases/biagram/` implements a character-level bigram language model using a `Brain(27 → 27)` with `ActivationType.None` followed by a manual softmax. Parameters can be saved/loaded via `SaveParameters` / `LoadParameters` to `parameters.txt`.
+
+`code/use-cases/playground/` is a scratch area for manual experiments.
+
+### Test structure
+
+Tests use NUnit 3 + Shouldly assertions. `MLPTrainingTests.Training` is a full end-to-end gradient-descent loop that runs until loss < 0.00001 or 10 000 steps — it is intentionally slow and exercises the entire autograd pipeline.

LICENSE

@@ -1,21 +1,21 @@
-MIT License
-
-Copyright (c) 2025 Franco Melandri
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.
+MIT License
+
+Copyright (c) 2025 Franco Melandri
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

README.md

@@ -1,10 +1,10 @@
-# tiny-brain
-
-The aim of the project is to create a small system based on neural network.
-
-This system is called `tiny brain`.
-
-I would like to consider this like a small brain that could be integrated inside bigger application
-I would like to use the brain as a base engine to perform tasks
-
-
+# tiny-brain
+
+The aim of the project is to create a small system based on neural network.
+
+This system is called `tiny brain`.
+
+I would like to consider this like a small brain that could be integrated inside bigger application
+I would like to use the brain as a base engine to perform tasks
+
+