INDEX

Explanations

phrases indicating an action that is not the primary focus or that the speaker is considering but hasn't fully committed to

oai_token-act-pair · gpt-3.5-turbo

phrases and structures that introduce or transition between thoughts or statements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

hov

-0.78

Leaks

-0.68

vation

-0.65

aft

-0.64

irie

-0.63

rance

-0.61

ivo

-0.60

rie

-0.60

HOU

-0.59

ãĥīãĥ©

-0.59

POSITIVE LOGITS

but

1.10

 haha

1.03

 because

0.97

 although

0.96

 lest

0.92

 though

0.92

so

0.91

 sorry

0.91

 since

0.89

 considering

0.89

Activations Density 0.241%

phrases indicating an action that is not the primary focus or that the speaker is considering but hasn't fully committed to

phrases and structures that introduce or transition between thoughts or statements

No Comments

No Known Activations