Downloads · 30 days
8
24% of all-time downloads
Llama-3.1-8B-Instruct-InstructionFollowing-SpectralSurgery-HNS8p2
tianzl66/Llama-3.1-8B-Instruct-InstructionFollowing-SpectralSurgery-HNS8p2
Llama-3.1-8B-Instruct-InstructionFollowing-SpectralSurgery-HNS8p2 is a machine learning model from tianzl66. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for peft.
This repository contains a Spectral Surgery adapter derived from the instruction-following LoRA checkpoint of Llama-3.1-8B-Instruct.
All-time downloads
33
Public
Repo size
168 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors168 MB · 100%
At a glance
- Library
- peft
- Access
- Public
- Created
- Aug 27, 2026
- Updated
- Aug 27, 2026
- SHA
- 0b61bb61
Base models
Model card
From the Hugging Face model README
Llama-3.1-8B-Instruct + Instruction Following + Spectral Surgery
This repository contains a Spectral Surgery adapter derived from the instruction-following LoRA checkpoint of Llama-3.1-8B-Instruct.
Base Model
meta-llama/Llama-3.1-8B-Instruct
Source LoRA
- Sequence length: 1024
- Global batch size: 128
- LoRA rank: 16
- Seed: 42
Spectral Surgery
- Fast HNS steps: 8
- Stable HNS steps: 2
Evaluation
Evaluation on IFEval.
| Model | Prompt Strict | Prompt Loose | Instruction Strict | Instruction Loose |
|---|---|---|---|---|
| Base | 65.80% | 72.83% | 75.54% | 81.18% |
| LoRA SFT | 65.99% | 70.98% | 74.82% | 79.26% |
| Spectral Surgery HNS 8+2 | 69.50% | 73.94% | 77.94% | 82.13% |
Relative to the vanilla LoRA checkpoint, Spectral Surgery HNS 8+2 improves:
- Prompt Strict by 3.51 percentage points
- Prompt Loose by 2.96 percentage points
- Instruction Strict by 3.12 percentage points
- Instruction Loose by 2.87 percentage points
The edited adapter also exceeds the base model on all four IFEval metrics.
- Library
- peft
- Created
- Aug 27, 2026
- Updated
- Aug 27, 2026