INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

essler

-0.81

hene

-0.80

cker

-0.79

yip

-0.78

enegger

-0.77

tek

-0.72

ofer

-0.72

omsky

-0.69

ozo

-0.69

nyder

-0.68

POSITIVE LOGITS

 withd

0.68

Fal

0.63

roll

0.62

 caught

0.62

 broom

0.59

wed

0.56

 across

0.56

eline

0.56

EVA

0.55

 hero

0.55

Activations Density 0.000%

No Known Activations

This feature has no known activations.