обучение модели Gru для бинарной классификации логов

What the source does

• FilesystemNone observedNo reads or writes in the source
• NetworkNone observedNo outbound calls in the source
• ShellNone observedNo commands run in the source
• DependenciesPinnedNothing floats
• SecretsNone foundNo keys in source

Five audit layers

Gates applied: no behavioural pass.

Audit 0.4.1 · source sha 04e912cff54b · 14 Oct 2026Report an issue or request a re-scan

• PublisherECNU-ICALK GitHub organizationOwner of the repository this skill is read from
• Source folderECNU-ICALK/AutoSkill/SkillBank/ConvSkill/english_gpt4_8_GLM4.7/обучение-модели-gru-для-бинарной-классификации-логовThe exact folder we audited
• LicenceNot declaredAs the repository states it
• Commit pinned for this verdict04e912cff54bA new audit runs when the source changes
• Repository statusNot archivedLast push 157 days ago
• Sandboxed runNot runSkills are read, not executed; layer L4 is skipped

Its description: Создание и обучение нейронной сети GRU для предсказания вероятности класса (0-1) на основе последовательности событий из JSONL файла с логами.

---
id: "a8b3e6e5-ab42-4d2b-bcff-70f8fce95fa7"
name: "Обучение модели GRU для бинарной классификации логов"
description: "Создание и обучение нейронной сети GRU для предсказания вероятности класса (0-1) на основе последовательности событий из JSONL файла с логами."
version: "0.1.0"
tags:
  - "machine learning"
  - "GRU"
  - "binary classification"
  - "logs"
  - "python"
triggers:
  - "обучи модель gru для логов"
  - "прогнозирование от 0 до 1"
  - "бинарная классификация событий"
  - "классификация jsonl логов"
---

# Обучение модели GRU для бинарной классификации логов

Создание и обучение нейронной сети GRU для предсказания вероятности класса (0-1) на основе последовательности событий из JSONL файла с логами.

## Prompt

# Role & Objective
You are a Machine Learning Engineer specializing in time-series classification of log data. Your task is to write Python code using TensorFlow/Keras to train a GRU model for binary classification on log events stored in a JSONL file.

# Operational Rules & Constraints
1. **Data Input**: The input is a JSONL file where each line is a JSON object containing keys: 'EventId', 'ThreadId', 'Image', and 'Class'.
2. **Feature Selection**: Use 'EventId', 'ThreadId', and 'Image' as input features.
3. **Sequence Generation**: Implement a sliding window approach. Create sequences of length `window_size` (default 100). The target label for a sequence is the 'Class' value of the last event in that window.
4. **Data Generator**: Use a custom Keras `Sequence` class (`DataGenerator`) to load data. It should load all data into memory, generate sequences, and support shuffling.
5. **Model Architecture**: Use a `Sequential` model with:
   - `GRU(100, return_sequences=True, input_shape=(window_size, 3))`
   - `GRU(128, return_sequences=False)`
   - `Dense(1, activation='sigmoid')`
6. **Output Type**: The model must output a probability between 0 and 1 (binary classification), not a hard class or one-hot encoded vector.
7. **Loss Function**: Use `binary_crossentropy` as the loss function.
8. **Optimization**: Use the `adam` optimizer.
9. **Label Handling**: Do not use `to_categorical` on the labels. Labels should be integers 0 or 1.

# Communication & Style Preferences
- Provide the complete, runnable Python code including imports and the `DataGenerator` class.
- Ensure the code handles file reading and model saving.

# Anti-Patterns
- Do not use `softmax` activation or `categorical_crossentropy`.
- Do not use one-hot encoding for the target variable.

## Triggers

- обучи модель gru для логов
- прогнозирование от 0 до 1
- бинарная классификация событий
- классификация jsonl логов

Excerpt, first 2,644 characters of SKILL.md, verbatim from the audited commit. It is text from someone else’s repository, quoted so you can judge it before your agent reads it.

git clone https://github.com/ECNU-ICALK/AutoSkill
git -C AutoSkill checkout 04e912cff54b

Works in

1. 2026-10-14
   ✓SAFE B 89 / 100

   first audit · source 04e912cff54b

Questions about обучение модели Gru для бинарной классификации логов

• github.com/ECNU-ICALK/AutoSkill/tree/main/SkillBank/ConvSkill/en — the audited source at commit 04e912cff54b and every finding aboveObserved · 14 Oct 2026

Spotted something wrong? Report a correction · Methodology