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

hower

-0.73

 harvesting

-0.66

 raining

-0.65

 suck

-0.63

 sucking

-0.63

 tails

-0.63

 Pool

-0.63

dow

-0.62

kson

-0.62

ulhu

-0.62

POSITIVE LOGITS

uten

0.68

ificant

0.67

--------------------------------

0.67

Effective

0.66

ocard

0.65

)\

0.64

atus

0.64

\\\\\\\\

0.64

Jun

0.64

================================================================

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.