INDEX

Explanations

phrases associated with options, comparisons, conditions, or consequences

oai_token-act-pair · gpt-3.5-turbo

terms and phrases indicating conditionality or alternatives

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

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

 guiActiveUnfocused

-0.76

 Heads

-0.76

BuyableInstoreAndOnline

-0.72

 Guards

-0.71

 Occupations

-0.69

 Gleaming

-0.67

 Norn

-0.65

 Recording

-0.64

lif

-0.64

 Picks

-0.64

POSITIVE LOGITS

rouse

1.15

venge

0.94

apply

0.89

ounce

0.86

ounced

0.82

gered

0.81

idate

0.81

andom

0.81

chev

0.76

apolog

0.76

Activations Density 0.466%

phrases associated with options, comparisons, conditions, or consequences

terms and phrases indicating conditionality or alternatives

No Comments

No Known Activations