J

Japanese Novel Reward Modernbert Ja 310m

Developed by Aratako
A reward model for Japanese novel quality assessment fine-tuned from modernbert-ja-310m, used to predict user evaluations of novel texts.
Downloads 15
Release Time : 3/3/2025

Model Overview

This model is primarily used to assess the quality of Japanese novel texts, suitable for scenarios like reinforcement learning in novel generation models. It indirectly evaluates text quality by predicting user evaluations through regression.

Model Features

Long text processing capability
Supports text inputs up to 8192 tokens in length, suitable for processing long novel content
Quality assessment
Evaluates novel text quality by predicting user evaluations, with output values ranging from 0-10
Reinforcement learning support
Optimized specifically for reinforcement learning scenarios in novel generation models

Model Capabilities

Japanese text understanding
Novel quality assessment
Long text processing
Regression prediction

Use Cases

Text generation optimization
Reinforcement learning for novel generation models
Used as a reward model for training novel generation AI
Provides quality scores on a 0-10 scale
Content quality assessment
Automatic novel quality scoring
Assesses quality of generated or human-written novels
Achieves Pearson correlation coefficient of 0.64 and Spearman correlation coefficient of 0.63
Featured Recommended AI Models
ยฉ 2025AIbase