J

Japanese Novel Reward Sarashina2.1 1b

Developed by Aratako
A Japanese novel quality assessment model fine-tuned from sbintuitions/sarashina2.1-1b, used to predict user evaluations of novel texts.
Downloads 40
Release Time : 2/25/2025

Model Overview

This model predicts user evaluations of input Japanese novel texts through regression, indirectly assessing text quality, primarily designed for applications like reinforcement learning in novel generation models.

Model Features

Long Text Processing Capability
Supports input texts up to 8192 tokens in length, suitable for processing complete novel segments.
Quality Score Prediction
Outputs scores in the 0-10 range to predict user evaluations of novel texts.
Reinforcement Learning Support
Specifically designed for reinforcement learning training scenarios in novel generation models.

Model Capabilities

Japanese Novel Quality Assessment
Text Regression Prediction
Long Text Processing

Use Cases

Natural Language Processing
Novel Generation Model Optimization
Used as a reward model to optimize the output quality of Japanese novel generation models
Provides quality score feedback on a 0-10 scale
Content Quality Filtering
Automatically filters high-quality novel text segments
Filters low-quality content based on score thresholds
Featured Recommended AI Models
ยฉ 2025AIbase