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

VPN

-0.79

ilde

-0.78

clave

-0.73

lang

-0.73

sg

-0.68

~~~~

-0.66

ËĪ

-0.66

SHIP

-0.66

^^^^

-0.65

DH

-0.65

POSITIVE LOGITS

 corrid

0.72

 calf

0.69

iott

0.67

 tiss

0.65

 balcony

0.64

iners

0.64

uder

0.64

 Yard

0.62

 Rapt

0.62

 topping

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.