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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 abduction

-0.78

UTE

-0.75

 repet

-0.68

 recess

-0.67

 deviation

-0.67

Scene

-0.65

Transaction

-0.64

 shootout

-0.62

 regression

-0.62

 rout

-0.62

POSITIVE LOGITS

 Hamb

0.76

 Cosponsors

0.74

iour

0.73

 Frie

0.68

hers

0.67

hered

0.66

aque

0.66

bos

0.65

Nit

0.64

imil

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.