Beaver-7B-v3.0 Open-Source Large Language Model - Focusing on Securely Utilizing Human Feedback, Free Deployment

Beaver 7b V3.0 GGUF

Developed by mradermacher

Beaver-7B-v3.0 is a 7B-parameter large language model based on the LLaMA architecture, focusing on safety and human feedback reinforcement learning (RLHF).

Large Language Model English#Safe Reinforcement Learning #Multiple Quantization Versions #Low-Resource Deployment

Downloads 405

Release Time : 4/1/2025

Model Overview

This model is trained on the PKU-SafeRLHF dataset, aiming to provide safe and reliable text generation capabilities for applications requiring high safety standards.

Model Features

Enhanced Safety

Trained on the PKU-SafeRLHF dataset, the model prioritizes safety and reliability in content generation.

Quantization Support

Offers multiple quantization versions (e.g., Q2_K, Q4_K_S, etc.), suitable for different hardware environments and performance needs.

Efficient Inference

Supports DeepSpeed optimization to improve inference efficiency.

Model Capabilities

Text Generation

Safe Content Filtering

Multi-turn Dialogue

Use Cases

Safe Dialogue Systems

Customer Service Bot

Used in customer service scenarios requiring high safety standards to ensure generated content complies with safety norms.

Content Moderation

Automated Content Filtering

Automatically detects and filters unsafe or inappropriate content.

🚀 Beaver-7B-v3.0 Static Quantization

This project provides static quantizations of the PKU-Alignment/beaver-7b-v3.0 model. It offers various quantized versions of the model in the GGUF format, which can be used with the transformers library.

📦 Model Information

Property	Details
Base Model	PKU-Alignment/beaver-7b-v3.0
Datasets	PKU-Alignment/PKU-SafeRLHF
Language	en
Library Name	transformers
Quantized By	mradermacher
Tags	reinforcement-learning-from-human-feedback, reinforcement-learning, rlhf, safety, ai-safety, deepspeed, beaver, llama, alpaca

📚 About

These are static quants of PKU-Alignment/beaver-7b-v3.0. At present, weighted/imatrix quants seem not to be available (by me). If they do not show up about a week after the static ones, I probably haven't planned for them. Feel free to request them by opening a Community Discussion.

💻 Usage

If you're unsure how to use GGUF files, refer to TheBloke's READMEs for more details, including how to concatenate multi - part files.

📋 Provided Quants

The following table lists the provided quantized models, sorted by size (not necessarily quality. IQ - quants are often preferable over similar sized non - IQ quants).

Link	Type	Size/GB	Notes
GGUF	Q2_K	2.6
GGUF	Q3_K_S	3.0
GGUF	Q3_K_M	3.4	lower quality
GGUF	Q3_K_L	3.7
GGUF	IQ4_XS	3.7
GGUF	Q4_K_S	4.0	fast, recommended
GGUF	Q4_K_M	4.2	fast, recommended
GGUF	Q5_K_S	4.8
GGUF	Q5_K_M	4.9
GGUF	Q6_K	5.6	very good quality
GGUF	Q8_0	7.3	fast, best quality
GGUF	f16	13.6	16 bpw, overkill

Here is a useful graph by ikawrakow comparing some lower - quality quant types (lower is better):

And here are Artefact2's thoughts on the matter: Artefact2's Gist

❓ FAQ / Model Request

Check Model Requests for answers to your questions and to request quantization of other models.

🙏 Thanks

I'm grateful to my company, nethype GmbH, for allowing me to use its servers and upgrading my workstation, enabling me to do this work in my free time.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご