alnrg2arg/test_wanda_240109

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:10.7BQuant:FP8Ctx Length:4kPublished:Jan 9, 2024License:cc-by-nc-4.0Architecture:Transformer Open Weights Warm

alnrg2arg/test_wanda_240109 is a 10.7 billion parameter language model, a pruned version of alnrg2arg/test. It is based on a combination of jeonsworld/CarbonVillain-en-10.7B-v2 and kyujinpy/Sakura-SOLAR-Instruct-DPO-v2, featuring a sparsity of 0.49. This model is designed for general language tasks, leveraging its pruned architecture for potentially more efficient inference.

Loading preview...

Model Overview

alnrg2arg/test_wanda_240109 is a 10.7 billion parameter language model derived from a pruning process applied to the alnrg2arg/test model. It integrates characteristics from two base models: jeonsworld/CarbonVillain-en-10.7B-v2 and kyujinpy/Sakura-SOLAR-Instruct-DPO-v2.

Key Characteristics

  • Parameter Count: 10.7 billion parameters.
  • Sparsity: The model exhibits a sparsity of 0.49, indicating that a significant portion of its parameters have been pruned. This often leads to reduced memory footprint and faster inference times compared to its dense counterparts, while aiming to retain performance.
  • Base Models: Built upon two distinct base models, suggesting a potential blend of their respective strengths and capabilities.

Good For

This model is suitable for users looking for a language model with a moderate parameter count that might offer efficiency benefits due to its pruned nature. Its foundation on established base models suggests applicability across a range of general language understanding and generation tasks.

Popular Sampler Settings

Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.

temperature
top_p
top_k
frequency_penalty
presence_penalty
repetition_penalty
min_p