AbacusResearch/haLLawa4-7b

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kPublished:Feb 19, 2024License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

The AbacusResearch/haLLawa4-7b is a 7 billion parameter language model, created by AbacusResearch, formed by merging mlabonne/Monarch-7B, paulml/OGNO-7B, and AbacusResearch/haLLAwa3 using the DARE TIES method. This model demonstrates strong general reasoning capabilities, achieving an average score of 75.25 on the Open LLM Leaderboard, with notable performance in commonsense reasoning and mathematical tasks. It is designed for applications requiring robust language understanding and generation across various benchmarks.

Loading preview...

haLLawa4-7b Model Overview

AbacusResearch/haLLawa4-7b is a 7 billion parameter language model developed by AbacusResearch. It is a product of a strategic merge using the DARE TIES method, combining the strengths of three distinct models: mlabonne/Monarch-7B, paulml/OGNO-7B, and AbacusResearch/haLLAwa3.

Key Capabilities & Performance

This model exhibits strong performance across a range of benchmarks, as evaluated on the Open LLM Leaderboard. Its average score is 75.25, indicating solid general-purpose language understanding and generation. Specific benchmark results include:

  • AI2 Reasoning Challenge (25-Shot): 71.50
  • HellaSwag (10-Shot): 88.36
  • MMLU (5-Shot): 64.49
  • TruthfulQA (0-shot): 74.27
  • Winogrande (5-shot): 82.40
  • GSM8k (5-shot): 70.51

These scores highlight its proficiency in commonsense reasoning, factual recall, and mathematical problem-solving.

Good For

  • General-purpose text generation and understanding tasks.
  • Applications requiring robust reasoning and problem-solving abilities.
  • Scenarios where a balanced performance across various benchmarks is desired.