Upwork is hiring a Language Model Engineer for Conversational Chatbot

Language Model Specialist for Enhancing Conversational Chatbot Responses

Job Description:

We are seeking a skilled Language Model (LLM) Specialist to join our team in developing a more conversational and engaging chatbot. The ideal candidate will have experience in working with large language models (LLMs) and a passion for creating human-like, chatty interactions in chatbot technology. This role involves leveraging existing LLMs to produce outputs that meet our "Metrics for Chattiness," aiming to make user interactions with our bot more memorable and enjoyable.

Responsibilities:

Emotion Classification:

Utilize open-source emotion classifiers (j-hartmann/emotion-english-distilroberta-base) to capture the emotion of the conversation so far.

Integrate this classification to guide the chatbot in generating emotion-aligned responses.

Initial Response Generation:

Generate a basic response using an open-source model fine-tuned on conversational datasets, such as Facebook's BlenderBot or Microsoft's DialoGPT.

Input parameters will include the user's question, relevant conversation excerpts, selected quotes, a history of the conversation, detected emotions, and guidelines for relevance and engagement.

3. Response Refinement through Multiple LLM Calls:

Employ Facebook's LLAMA - 7b or a similar general and stable LLM for refining responses through multiple calls.

First Call: Simplify the language by reducing word count, sentence number, and sentence length.

Second Call: Enhance empathy/sentiment and increase lexical diversity in the response.

Final Call: Incorporate personal pronouns, relevant questions, and exclamations to boost engagement.