INDEX

Explanations

proper nouns or names related to entities like people, places, organizations, or events

oai_token-act-pair · gpt-3.5-turbo

instances of the letter 'A' in the text

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

 envy

-0.68

ographs

-0.66

 himself

-0.64

mods

-0.62

Â·Â·

-0.61

falls

-0.59

faces

-0.58

dn

-0.58

},{"

-0.58

ometers

-0.58

POSITIVE LOGITS

cknowled

1.07

usterity

1.03

UTH

1.00

cknow

0.98

UG

0.97

gency

0.94

ntil

0.94

ircraft

0.93

verages

0.93

uckland

0.89

Activations Density 0.085%

proper nouns or names related to entities like people, places, organizations, or events

instances of the letter 'A' in the text

No Comments

No Known Activations