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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 curtains

-0.75

 overd

-0.73

 insulation

-0.73

 owed

-0.73

essential

-0.69

 Essential

-0.68

Ess

-0.65

 blindness

-0.65

osc

-0.62

AUD

-0.62

POSITIVE LOGITS

tumblr

1.00

0.74

Tumblr

0.73

eele

0.71

twitch

0.70

Reddit

0.70

psons

0.69

zo

0.67

irect

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.