S

Splade Disco Human

Developed by slupart
A conversational search version adapted from the SPLADE++ model, fine-tuning the query encoder on the QReCC dataset to optimize multi-turn conversational search performance.
Downloads 22
Release Time : 4/17/2025

Model Overview

This model is an adapted version of SPLADE++ for conversational search, retaining the original document encoder while fine-tuning the query encoder on the QReCC dataset, specifically designed for multi-turn conversational search. Through knowledge distillation of human-rewritten texts, the model better captures the semantics of conversational queries.

Model Features

Conversational Search Optimization
Fine-tunes the query encoder on the QReCC dataset, specifically optimizing performance for multi-turn conversational search.
Knowledge Distillation
Through knowledge distillation of human-rewritten texts, the model better captures the semantics of conversational queries.
Asymmetric Architecture
Adopts SPLADE's asymmetric architecture, with independent models for query and document representations.

Model Capabilities

Multi-turn Conversational Search
Query Expansion
Document Expansion
Passage Retrieval

Use Cases

Information Retrieval
Multi-turn Conversational Search System
Used to build search systems that understand conversational context and return relevant documents.
The optimized query encoder better understands semantic changes in conversations.
Intelligent Customer Service
Applied in customer service systems to help understand users' multi-turn questions and provide accurate answers.
Understands user intent through conversation history, improving answer accuracy.
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase