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

PDATE

-0.82

-0.72

ãĤ·ãĥ£

-0.71

sites

-0.70

ãĥĭ

-0.68

night

-0.67

Spread

-0.67

 bookmark

-0.66

 simul

-0.66

Redditor

-0.65

POSITIVE LOGITS

icide

0.70

iesel

0.68

arov

0.66

oys

0.63

onomic

0.63

inton

0.63

icides

0.61

ocese

0.60

 Titanium

0.60

uclear

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.