INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 vocals

-0.71

 lyrics

-0.67

 Logic

-0.67

 Sorceress

-0.67

 preacher

-0.65

 immersion

-0.65

 mechanics

-0.63

 jams

-0.63

 grav

-0.63

 hacking

-0.63

POSITIVE LOGITS

posted

0.88

ante

0.78

ays

0.78

acqu

0.78

gallery

0.75

 ILCS

0.73

hou

0.72

toggle

0.72

boa

0.71

abe

0.71

Activations Density 0.000%

No Known Activations

This feature has no known activations.