INDEX

Explanations

phrases related to actions taken by someone or something

oai_token-act-pair · gpt-3.5-turbo

instances of the verb "had" indicating past actions or experiences

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

hammer

-0.66

bie

-0.65

hack

-0.64

PI

-0.63

bies

-0.58

 Voters

-0.57

etting

-0.57

atin

-0.56

externalActionCode

-0.56

porting

-0.56

POSITIVE LOGITS

 been

1.27

 begun

1.13

 undergone

1.08

 gone

1.06

 gotten

1.03

iths

1.02

been

0.92

 become

0.90

 previously

0.88

 fallen

0.87

Activations Density 0.177%

phrases related to actions taken by someone or something

instances of the verb "had" indicating past actions or experiences

No Comments

No Known Activations