INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

heastern

-0.73

sonian

-0.73

idel

-0.69

eton

-0.69

bilt

-0.68

 convol

-0.68

uman

-0.66

Mah

-0.64

zees

-0.63

onda

-0.62

POSITIVE LOGITS

 INTER

0.71

imity

0.70

trap

0.67

 procedures

0.66

 rapport

0.65

 arrangements

0.65

 oper

0.63

 polic

0.60

 Committees

0.59

%:

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.