INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

velop

-0.87

nces

-0.81

igslist

-0.79

Gum

-0.73

glers

-0.67

gow

-0.66

ewater

-0.65

FP

-0.62

gum

-0.62

 Debt

-0.59

POSITIVE LOGITS

iour

0.87

iron

0.75

chlor

0.71

morning

0.69

hent

0.68

chest

0.68

TN

0.67

POSE

0.67

advertisement

0.65

Redditor

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.