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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ĸļ

-0.79

 Ashe

-0.77

SHIP

-0.70

estern

-0.68

Geh

-0.68

 Glow

-0.66

stretched

-0.66

 flanked

-0.66

 Solitaire

-0.64

 Shroud

-0.64

POSITIVE LOGITS

 agree

1.64

 agrees

1.07

 disagree

1.00

agree

0.92

 agreed

0.91

 agreeing

0.78

 disagrees

0.68

ids

0.68

rak

0.68

acists

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.