Update README.md

README.md

@@ -69,4 +69,40 @@ nlp = pipeline("text-classification", model=model_name)
 # Example usage
 result = nlp("Example shell command or exploit input")
 print(result)
-```
+```
+
+## Training Details
+### Training Data
+The model was fine-tuned on the following datasets:
+
+- Canstralian/ShellCommands: A collection of shell commands used in cybersecurity contexts.
+- Canstralian/CyberExploitDB: A curated set of known exploits and vulnerabilities.
+Further details on the preprocessing of these datasets can be found in their respective dataset cards.
+
+## Training Procedure
+### Preprocessing
+The data was preprocessed to remove any sensitive or personally identifiable information. Text normalization and tokenization were applied to ensure consistency across the datasets.
+
+### Training Hyperparameters
+Training regime: fp16 mixed precision
+Evaluation
+Testing Data, Factors & Metrics
+Testing was performed on both synthetic and real-world shell command and exploit datasets, focusing on their ability to correctly parse shell commands and identify exploit signatures.
+
+## Factors
+The evaluation factors included:
+
+Model performance across different types of shell commands and exploits.
+Accuracy, precision, recall, and F1-score in detecting known exploits.
+## Metrics
+Metrics used for evaluation include:
+
+- Accuracy: Percentage of correct predictions made by the model.
+- Precision: The number of relevant instances among the retrieved instances.
+- Recall: The number of relevant instances that were retrieved.
+- F1-score: The harmonic mean of precision and recall.
+## Results
+The model performs well on standard shell command parsing tasks and exploit detection, with high accuracy for common exploits. However, its performance may degrade on newer or less common exploits.
+
+## Summary
+The model is well-suited for cybersecurity applications involving shell command and exploit detection. While it excels in these areas, users should monitor its performance for emerging threats and unusual attack patterns.