---
base_model: facebook/bart-large-mnli
datasets:
- reddgr/nli-chatbot-prompt-categorization
library_name: transformers
license: mit
tags:
- generated_from_keras_callback
model-index:
- name: zero-shot-prompt-classifier-bart-ft
  results: []
---

# zero-shot-prompt-classifier-bart-ft

This model is a fine-tuned version of [facebook/bart-large-mnli](https://huggingface.co/facebook/bart-large-mnli) on the [reddgr/nli-chatbot-prompt-categorization](https://huggingface.co/datasets/reddgr/nli-chatbot-prompt-categorization) dataset.

The purpose of the model is to help classify chatbot prompts into categories that are relevant in the context of working with LLM conversational tools: 
coding assistance, language assistance, role play, creative writing, general knowledge questions... 

The model is fine-tuned and tested on the natural language inference (NLI) dataset [reddgr/nli-chatbot-prompt-categorization](https://huggingface.co/datasets/reddgr/nli-chatbot-prompt-categorization)

Below is a confusion matrix calculated on zero-shot inferences for the 10 most popular categories in the Test split of [reddgr/nli-chatbot-prompt-categorization](https://huggingface.co/datasets/reddgr/nli-chatbot-prompt-categorization) at the time of the first model upload. The classification with the base model on the same small test dataset is shown for comparison:

![Zero-shot prompt classification confusion matrix for reddgr/zero-shot-prompt-classifier-bart-ft](https://talkingtochatbots.com/wp-content/uploads/2024/12/zero-shot-prompt-classification-comparison-57-accuracy.png)

The current version of the fine-tuned model outperforms the base model [facebook/bart-large-mnli](https://huggingface.co/facebook/bart-large-mnli) by 23 percentage points (57% accuracy vs 34% accuracy) in a test set with 10 candidate zero-shot classes (the most frequent categories in the test split of [reddgr/nli-chatbot-prompt-categorization](https://huggingface.co/datasets/reddgr/nli-chatbot-prompt-categorization)).

The chart below compares the results for the 12 most popular candidate classes in the Test split, where the base model's zero-shot accuracy is outperformed by 25 percentage points:

![Zero-shot prompt classification confusion matrix for reddgr/zero-shot-prompt-classifier-bart-ft](https://talkingtochatbots.com/wp-content/uploads/2024/12/zero-shot-prompt-classification-comparison-12-classes-56-accuracy.png)

The dataset and the model are continously updated as they assist with content publishing on my website [Talking to Chatbots](https://talkingtochatbots) 


## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': None, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': False, 'is_legacy_optimizer': False, 'learning_rate': 5e-06, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-07, 'amsgrad': False}
- training_precision: float32

### Training results

| Train Loss | Train Accuracy | Validation Loss | Validation Accuracy | Epoch |
|:----------:|:--------------:|:---------------:|:-------------------:|:-----:|
| 0.9969     | 0.5490         | 0.9182          | 0.6225              | 0     |
| 0.7647     | 0.6601         | 1.0025          | 0.5441              | 1     |
| 0.6465     | 0.7157         | 1.1472          | 0.5392              | 2     |
| 0.5849     | 0.7418         | 1.1974          | 0.5049              | 3     |
| 0.4474     | 0.7843         | 1.5942          | 0.4657              | 4     |


### Framework versions

- Transformers 4.44.2
- TensorFlow 2.18.0-dev20240717
- Datasets 2.21.0
- Tokenizers 0.19.1