INDEX

Explanations

mentions of specific words or phrases related to "box"

oai_token-act-pair · gpt-3.5-turbo

references to box office performance

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ufact

-0.83

 Hots

-0.78

uates

-0.76

uating

-0.74

uated

-0.74

vironment

-0.70

baugh

-0.69

 GOODMAN

-0.69

 Lauder

-0.69

uate

-0.68

POSITIVE LOGITS

box

1.03

boxes

1.03

cars

0.99

buster

0.96

ers

0.95

sets

0.93

 boxes

0.90

nut

0.89

door

0.89

car

0.87

Activations Density 0.033%

mentions of specific words or phrases related to "box"

references to box office performance

No Comments

No Known Activations