INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Honolulu

-0.77

 Wichita

-0.69

 Countdown

-0.69

 Caucasus

-0.65

KS

-0.64

 Cato

-0.62

 Knox

-0.61

live

-0.61

 Hawai

-0.59

 Okinawa

-0.59

POSITIVE LOGITS

 reconc

0.78

BILITIES

0.75

 metic

0.73

BuyableInstoreAndOnline

0.70

 depreciation

0.70

 disadvant

0.68

pard

0.68

 rehe

0.68

robe

0.66

 thous

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.