INDEX

Explanations

numbers and quantities that have been totaled or summed up

oai_token-act-pair · gpt-3.5-turbo

phrases related to numerical totals or aggregates

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

 behaved

-0.68

 boun

-0.65

scape

-0.63

bies

-0.62

generic

-0.60

 assail

-0.60

 Advertisement

-0.60

 hire

-0.59

rir

-0.59

croft

-0.58

POSITIVE LOGITS

 totals

1.03

 totaled

0.95

 totaling

0.84

ources

0.81

izational

0.79

urgical

0.75

mits

0.74

OGR

0.73

aurus

0.72

 digits

0.72

Activations Density 0.015%

numbers and quantities that have been totaled or summed up

phrases related to numerical totals or aggregates

No Comments

No Known Activations