INDEX

Explanations

phrases indicating the introduction of a new or additional concept or subject

oai_token-act-pair · gpt-3.5-turbo

phrases indicating repeated instances or patterns of behavior or events

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

isters

-0.78

alties

-0.77

hips

-0.73

lee

-0.72

kee

-0.71

liest

-0.69

=-=-=-=-

-0.68

Always

-0.68

oes

-0.67

Hos

-0.65

POSITIVE LOGITS

worldly

1.06

 notch

0.85

 contender

0.79

wcs

0.78

 avenue

0.77

natureconservancy

0.76

 dimension

0.76

 installment

0.76

judicial

0.76

 Flavoring

0.74

Activations Density 0.038%

phrases indicating the introduction of a new or additional concept or subject

phrases indicating repeated instances or patterns of behavior or events

No Comments

No Known Activations