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

Gal

-0.73

iator

-0.67

 Drag

-0.66

ihad

-0.65

 Whis

-0.65

LV

-0.65

++)

-0.65

 Giles

-0.64

Message

-0.63

 Ariel

-0.62

POSITIVE LOGITS

chwitz

0.94

©¶æ

0.77

boro

0.76

creen

0.70

owship

0.69

icrobial

0.67

reditary

0.67

afety

0.67

hower

0.66

cribed

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.