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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ographer

-0.72

busters

-0.70

Ĭ±

-0.70

obin

-0.68

osphere

-0.66

WAR

-0.65

School

-0.65

details

-0.63

stories

-0.63

HOW

-0.63

POSITIVE LOGITS

phabet

0.75

outheastern

0.75

tein

0.71

 triangular

0.65

 contiguous

0.63

rient

0.63

forward

0.63

 gelatin

0.62

ection

0.62

 ellipt

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.