INDEX

Explanations

phrases related to unexpected or surprising events

oai_token-act-pair · gpt-3.5-turbo

instances of imminent events or actions that involve seeing or perceiving something

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

tein

-0.75

ãĥĥ

-0.66

 sugg

-0.63

angan

-0.62

uster

-0.61

 basis

-0.60

 guaranteed

-0.59

 Wond

-0.58

este

-0.58

 supervised

-0.58

POSITIVE LOGITS

 unfold

0.84

 emerge

0.75

aily

0.75

escape

0.69

progress

0.69

 firsthand

0.68

amorph

0.68

 trending

0.66

soDeliveryDate

0.65

 pandemonium

0.64

Activations Density 0.534%

phrases related to unexpected or surprising events

instances of imminent events or actions that involve seeing or perceiving something

No Comments

No Known Activations