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

ractor

-0.75

etheus

-0.75

 Against

-0.68

 Kahn

-0.68

WAY

-0.68

NZ

-0.66

 Navajo

-0.65

Ĥİ

-0.64

 Sunder

-0.64

TYPE

-0.63

POSITIVE LOGITS

ashes

0.78

ares

0.73

 rare

0.71

iblings

0.68

 congrat

0.67

 Anniversary

0.65

liam

0.65

birth

0.64

ardy

0.64

chieve

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.