---
license: apache-2.0
---
# Bylastic: A Log Classifier Compatible with Elastic

## Introduction

**Bylastic** is an advanced AI-based log classification model specifically designed to categorize records into three levels: **ERROR**, **WARNING**, and **INFO**. Created by **Byviz Analytics**, this model is fully optimized for integration with Elastic, offering an efficient and accurate solution for log management and analysis.

## Key Features

- **Accurate Classification**: Bylastic classifies logs into three critical categories: **ERROR**, **WARNING**, and **INFO**, helping to quickly identify problems, warnings, and general system information.
- **Full Elastic Compatibility**: Designed to seamlessly integrate with Elastic, Bylastic facilitates data ingestion and analysis within the Elastic ecosystem.
- **High Performance**: Optimized to process large volumes of logs, ensuring fast and efficient performance even in high-demand environments.
- **Easy Integration**: Bylastic can be easily integrated into your existing log processing pipelines, reducing implementation time and associated costs.

### Precision Bylastic vs Bert
Precision in categorizing logs with example data, Bert cannot identify the categories
![image/png](https://cdn-uploads.huggingface.co/production/uploads/65bc645cb7db0ab095f10320/hpwBT4dKWt3rIKAjn4UM4.png)

## Requeriments Bylastic vs Bert
![image/png](https://cdn-uploads.huggingface.co/production/uploads/65bc645cb7db0ab095f10320/gpCmIOXPScF4R2LgMre0z.png)


## How It Works

Bylastic utilizes advanced natural language processing (NLP) techniques, a branch of artificial intelligence (AI), to analyze and categorize logs. The model has been trained with a diverse set of log data, ensuring high accuracy in classification.

### Log Categories

- **ERROR**: Logs indicating critical failures or serious problems that require immediate attention.
- **WARNING**: Logs indicating potential issues that could become errors if not properly managed.
- **INFO**: Informational logs that provide details about the normal functioning of the system.

## Integration with Elastic

Integrating Bylastic with Elastic is straightforward and direct. Here is a quick guide to integrate the model into your Elastic environment:

1. **Installation**: Download Bylastic from Hugging Face.
2. **Upload the Model**: Use `eland` to upload the model to your Elastic cluster.
3. **Create an inference pipeline in Elastic**
   
## Benefits

- **Improved Log Management**: Facilitates the identification and resolution of issues by automatically classifying logs.
- **Time Savings**: Reduces the time required to manually review and categorize logs.
- **Higher Accuracy**: Minimizes human errors in log classification.
- **Easy Integration**: Seamlessly integrates with Elastic, leveraging Elastic's advanced search and analysis capabilities.


# Load model directly

```python
from transformers import AutoTokenizer, AutoModelForSequenceClassification

tokenizer = AutoTokenizer.from_pretrained("byviz/bylastic_classification_logs")
model = AutoModelForSequenceClassification.from_pretrained("byviz/bylastic_classification_logs")
```


## Test with Elastic
![image/png](https://cdn-uploads.huggingface.co/production/uploads/65bc645cb7db0ab095f10320/tTaQ_H84CqPR0b3Sl0Thv.png)

# Contact
If you need AI models with personalized training and compatible with elastic or have any suggestions, you can contact:
- **ivan.frias@byviz.com**
- **ivan.frias@elastic.co**