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

gall

-0.71

cellent

-0.69

 bench

-0.69

uristic

-0.65

 evaluations

-0.64

igger

-0.62

Trend

-0.62

idious

-0.60

strous

-0.58

Mob

-0.58

POSITIVE LOGITS

 Seym

0.82

Downloadha

0.77

aiden

0.67

argon

0.66

 Directions

0.64

thodox

0.64

 chall

0.64

 Username

0.63

athered

0.63

 emancipation

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.