INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 verbally

-0.74

 disarm

-0.69

 strategically

-0.64

 internally

-0.64

ishly

-0.63

bil

-0.63

 financially

-0.62

 fired

-0.62

aith

-0.61

 physically

-0.61

POSITIVE LOGITS

interstitial

0.75

ukemia

0.74

Developer

0.73

UX

0.72

à¤¾

0.72

math

0.69

CLASSIFIED

0.68

iasm

0.67

cium

0.66

mi

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.