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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rontal

-0.80

 ItemLevel

-0.78

 Personality

-0.73

 Corrections

-0.73

 Communication

-0.71

 Perception

-0.71

iesta

-0.71

 Opinion

-0.65

 kcal

-0.65

vation

-0.64

POSITIVE LOGITS

isite

0.68

igs

0.68

 exhibited

0.67

akespe

0.67

igham

0.67

Sund

0.67

 Dresden

0.65

pei

0.65

uming

0.64

thur

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.