AIbase
EN
Home
Absolute Harmfulness Predictor Redteam Osst
A

Absolute Harmfulness Predictor Redteam Osst

Developed by safepaca
A model fine-tuned based on microsoft/deberta-v3-large for evaluation tasks. The specific task type is unknown.
Large Language Model
Transformers
Open Source License:MIT#Text harm assessment#High-precision fine-tuning#DeBERTa architecture
Downloads 114
Release Time : 9/25/2023

Model Overview

This model is a fine-tuned version of microsoft/deberta-v3-large, mainly used for evaluation tasks. The specific evaluation content and dataset information are not provided.

Model Features

Fine-tuned based on DeBERTa-v3-large
Use the powerful DeBERTa-v3-large as the base model for fine-tuning
Optimized for evaluation tasks
Optimized specifically for a certain evaluation task (specific task type not specified)
Relatively stable training performance
From the training results, the validation loss and mean squared error gradually decreased and stabilized during the training process

Model Capabilities

Text evaluation
Classification tasks

Use Cases

Content evaluation
Harmful content detection
May be used to evaluate the harmfulness of text content
Validation loss 1.1439, Mean squared error 1.1439
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
EnglishįŽ€äŊ“中文įšéĢ”中文ãĢãģんご
Š 2025AIbase