Jeesup/svd-safety-l3_remove20_swapgapiter_b010_r06

TEXT GENERATIONPricing:Input $0.37 / Cached $0.074 / Output $0.38Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:8kTool Calling:SupportedPublished:Sep 17, 2026License:llama3Architecture:Transformer Featherless Exclusive Cold

Jeesup/svd-safety-l3_remove20_swapgapiter_b010_r06 is an 8 billion parameter Llama-3-8B-Instruct checkpoint compressed with SVD-LLM, retaining 80.0% of its original parameters. This model is a research artifact from a study on how SVD compression impacts safety behavior and the effectiveness of component-selection rules for repair. It is specifically designed for measuring safety/utility trade-offs under compression, rather than serving as a general-purpose chat model.

Loading preview...

Model Overview

This model, svd-safety-l3_remove20_swapgapiter_b010_r06, is an 8 billion parameter Llama-3-8B-Instruct checkpoint that has undergone significant modification. It was initially compressed using SVD-LLM, resulting in the removal of 20.02% of its parameters, leaving 80.0% of the dense parameters. Following compression, the model was further edited through 6 out of 10 rounds of iterative parameter-neutral swaps, guided by the gap_iter selection rule.

Key Characteristics

  • Base Model: meta-llama/Meta-Llama-3-8B-Instruct
  • Compression Method: SVD-LLM, removing 20.02% of parameters.
  • Restoration/Editing: 6 rounds of iterative swaps using the gap_iter rule, restoring 6006 components and swapping out 6006 components.
  • Resulting Parameter Fraction: 0.7998 (approximately 80% of original dense parameters).
  • Research Focus: This model is a research artifact from a study investigating the impact of SVD compression on safety behavior and methods for repairing safety degradation.

Intended Use and Limitations

This checkpoint is not intended as a general-purpose chat model. Its primary purpose is to serve as an experimental subject for measuring safety/utility trade-offs under compression. The study deliberately includes arms that are safety-degraded relative to the base Llama-3-8B-Instruct model. Users should treat this model as an experimental subject and conduct their own evaluations before drawing conclusions or deploying it. The model is bound by the Meta Llama 3 Community License.

Finetunes
1 models