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

uckland

-0.90

iannopoulos

-0.86

tro

-0.80

negie

-0.77

enhagen

-0.77

encies

-0.76

borgh

-0.75

idate

-0.74

ussen

-0.72

oppy

-0.71

POSITIVE LOGITS

ene

0.67

 Prison

0.67

 Vault

0.67

 Compton

0.66

urized

0.66

 Codex

0.64

 Abyssal

0.63

 Assass

0.63

 Compact

0.62

 Blast

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.