Jeesup/svd-safety-l3_remove30_swapgapiter_b010_r04

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_remove30_swapgapiter_b010_r04 is an 8 billion parameter Llama-3-8B-Instruct checkpoint, compressed using SVD-LLM to 70% of its original parameters. This model is a research artifact from a study on how SVD compression impacts safety behavior and the effectiveness of iterative parameter-neutral swapping for repair. It is specifically designed for measuring safety/utility trade-offs under compression, rather than as a general-purpose chat model.

Loading preview...

Model Overview

This model, svd-safety-l3_remove30_swapgapiter_b010_r04, is a research artifact derived from meta-llama/Meta-Llama-3-8B-Instruct. It has undergone significant compression and iterative parameter swapping as part of a study to understand and mitigate the impact of compression on model safety.

Key Characteristics

  • Base Model: Meta-Llama-3-8B-Instruct.
  • Compression Method: SVD-LLM, resulting in 30.01% parameter removal, leaving 69.99% of the original parameters.
  • Safety Repair Mechanism: Utilizes the gap_iter selection rule for iterative parameter-neutral swapping, with 4 out of 10 planned rounds applied.
  • Restoration Budget: A total restore budget of 1.000% of dense parameters, with 0.100% applied per round.
  • Parameters Swapped: 27,893,760 parameters (0.40% of dense projection parameters) were swapped in.

Measured Performance (Safety Metrics)

As an experimental subject, this checkpoint exhibits specific safety-related metrics:

  • AdvBench ASR (HarmBench judge): 0.0000
  • StrongREJECT ASR (HarmBench judge): 0.0100
  • Macro over-refusal (WildGuard): 0.6519

Intended Use and Limitations

This model is not intended as a general-purpose deployable assistant. Its primary purpose is to serve as an experimental subject for measuring safety/utility trade-offs under compression. The study aims to quantify how compression alone can raise attack-success rates and test recovery methods. Users should treat this as a research checkpoint and conduct their own evaluations before drawing conclusions about its safety or utility.

Finetunes
1 models