INDEX

Explanations

phrases related to communication where information is shared or instructions are given

oai_token-act-pair · gpt-3.5-turbo

instances of the word "told."

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

 sidx

-0.84

adesh

-0.80

elsh

-0.73

imposed

-0.69

ccording

-0.68

ibaba

-0.66

arthy

-0.65

berus

-0.65

ikes

-0.64

everal

-0.64

POSITIVE LOGITS

tale

1.30

us

1.07

him

1.05

me

0.99

 them

0.94

ingly

0.92

 everyone

0.78

 stories

0.78

Ŀ

0.76

 jokes

0.74

Activations Density 0.069%

phrases related to communication where information is shared or instructions are given

instances of the word "told."

No Comments

No Known Activations