Chatbot Suite

Welcome to the Chatbot Suite project, a collection of diverse chatbots for various purposes, including question answering, language translation, and document-based interactions. Each chatbot leverages state-of-the-art language models and technologies to provide unique functionalities.

Directory Structure

The project is organized into several directories, each housing a distinct chatbot application:

1. QA_LLms: Implements a chatbot capable of question answering and language translation. Utilizes Google/Flax-T5-XXL for question answering and Helsinki-NLP/opus-mt-en-{target_language} for language translation.

2. chatbot_pdf: Contains a chatbot application for question answering from documents, such as PDFs. Utilizes Google/Flan-T5-XXLfrom Huggingface and employs faiss for embedding-related processes. The implementation is integrated into the langchain library.

3. llama2chatbot: Implements a chatbot using the llama2 pre-trained model. Offers enhanced text generation for question answering from documents. Utilizes Pinecone vector database for efficient processing.

4. ollama_models: Includes a chatbot for document-based question answering using the ollama class's fine-tuned model named orca-mini. Offers lower memory usage compared to llama2.

5. GenerativeAI: Utilizes Google's palm model to create a chatbot for various conversational purposes.

6. Bert-finetuned bot: Utilizes Bert pre-trained model for creating chatbot with a json dataset.

7. Openai_turbo_chatbot: Utilizes Openai turbo model for Retrival augmented generation from pdf with streaming response.

8. Llavaforimage: Utilizes Llava multimodal for generating answer from image when we ask a question with prompt engineering.

9. gpt4vision: Utilizes gpt4vision multimodal for retrival augmented generation for texts, tables , diagram summarization etc from images.

Chatbot Descriptions

1. QA_LLms:

   • Capabilities: Question answering, language translation.
   • Models: Google/Flax-T5-XXL, Helsinki-NLP/opus-mt-en-{target_language}.

2. pdf_chatbot:

   • Capabilities: Document-based question answering, PDF interaction.
   • Model: Google/Flan-T5-XXL.
   • Libraries: Langchain, faiss.

3. llama2chatbot:

   • Capabilities: Enhanced text generation for question answering.
   • Model: llama2 pre-trained model.
   • Database: Pinecone vector.

4. ollama_models:

   • Capabilities: Document-based question answering with low memory usage.
   • Model: orca-mini.

5. GenerativeAI:

   • Capabilities: Conversational chatbot using Google's palm model.

6. Bert fine-tuned Bot:

   • Capabilities: Conversational chatbot using Bert Pre-trained model.

7. Openai_turbo_chatbot:

   • Capabilities: Conversational chatbot using openai turbo model for streamed data augmented generation from documents.

8. Llavaforimage:

   • Capabilities: Conversational chatbot using Llava model from images.

9. gpt4vision:

   • Capabilities: Conversational chatbot using gpt4vision model from images.

Dependencies

The project relies on the following dependencies:

• Python 3.x
• Huggingface Transformers
• Langchain Library
• Faiss
• Pinecone

For specific dependencies of each chatbot, refer to the README file in the respective directory.

Contributing

We welcome contributions to enhance and expand the capabilities of our chatbots. Feel free to submit issues, feature requests, or pull requests.

Acknowledgments

We appreciate the contributions of the open-source community and the developers behind the libraries and models used in this project.

Research and Future Development

Actively researching new technologies and enhancements for future iterations. Stay tuned for updates and exciting developments in the field of conversational AI.