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

senal

-0.73

akings

-0.73

 inheritance

-0.71

cade

-0.71

 Strikes

-0.64

atana

-0.63

entious

-0.62

ahon

-0.62

ication

-0.62

FH

-0.61

POSITIVE LOGITS

youtu

0.70

EG

0.65

trak

0.65

arag

0.65

fred

0.64

Bra

0.63

pring

0.62

ISO

0.62

milo

0.62

EXT

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.