INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

mort

-0.83

¶ħ

-0.82

mble

-0.74

riger

-0.71

unts

-0.70

olls

-0.70

 limbs

-0.65

anson

-0.64

 nomine

-0.64

 Lumpur

-0.64

POSITIVE LOGITS

VIEW

0.75

owitz

0.74

owicz

0.71

=\"

0.70

Accessory

0.69

furt

0.69

 Ellison

0.66

NX

0.64

WAR

0.64

 Turing

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.