lxj285/Llama-3.1-70B-Instruct-abliterated

TEXT GENERATIONConcurrent Unit Cost:4Model Size:70BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jul 2, 2026License:llama3.1Architecture:Transformer Featherless Exclusive Cold

lxj285/Llama-3.1-70B-Instruct-abliterated is a 70 billion parameter instruction-tuned generative language model from Meta's Llama 3.1 family, optimized for multilingual dialogue use cases. It features a 32,768 token context length and is designed for commercial and research applications, outperforming many chat models on industry benchmarks. This version has layer 14 orthogonalized, indicating a specific architectural modification.

Loading preview...

Model Overview

lxj285/Llama-3.1-70B-Instruct-abliterated is a 70 billion parameter instruction-tuned model from Meta's Llama 3.1 series, featuring a 32,768 token context length. It is built on an optimized transformer architecture and fine-tuned using supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF) for helpfulness and safety. This specific model has undergone an orthogonalization of layer 14, a unique modification.

Key Capabilities

  • Multilingual Support: Optimized for dialogue in English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai, with potential for other languages through fine-tuning.
  • Extended Context Window: Supports a 128k token context length, enabling processing of longer inputs and generating more comprehensive responses.
  • Enhanced Performance: Outperforms many open-source and closed chat models on common industry benchmarks, particularly in general reasoning, code generation, and mathematical tasks.
  • Tool Use Integration: Supports multiple tool use formats, facilitating integration with external functions and services.
  • Commercial & Research Use: Intended for a wide range of commercial and research applications, including assistant-like chat and natural language generation.

Good For

  • Multilingual Chatbots: Developing conversational AI agents that can interact effectively across multiple supported languages.
  • Code Generation: Excelling in programming tasks, as indicated by strong performance on HumanEval and MBPP++ benchmarks.
  • Complex Reasoning: Handling intricate reasoning and mathematical problems, with high scores on benchmarks like MATH and GSM-8K.
  • Tool-Augmented Applications: Building AI systems that leverage external tools and APIs for enhanced functionality.
  • Research & Development: Exploring advanced LLM capabilities and fine-tuning for specific domain applications.

Finetunes
1 models