INDEX

Explanations

numerical identifiers or markers likely in a list or ranking context

oai_token-act-pair · gpt-3.5-turbo

numerical data or statistics, particularly related to voting outcomes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ipal

-0.76

ened

-0.73

yright

-0.70

 Examiner

-0.69

atre

-0.68

bear

-0.68

present

-0.68

morning

-0.68

stract

-0.68

uscript

-0.67

POSITIVE LOGITS

wcsstore

0.87

0.85

 msec

0.79

ESSION

0.73

0.72

UTF

0.69

 SHARES

0.69

 ILCS

0.68

mph

0.68

Activations Density 0.021%

numerical identifiers or markers likely in a list or ranking context

numerical data or statistics, particularly related to voting outcomes

No Comments

No Known Activations