INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 royalty

-0.66

 flow

-0.65

 deserve

-0.65

 Mubarak

-0.64

 syndrome

-0.62

 Rohingya

-0.61

gered

-0.60

yrics

-0.60

 permitting

-0.60

 breakdown

-0.59

POSITIVE LOGITS

swick

0.78

zos

0.77

ccoli

0.75

 CLIENT

0.71

oult

0.71

ricanes

0.70

kef

0.68

umbn

0.68

clave

0.68

ruary

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.