INDEX

Explanations

the word "whenever" followed by an action or event

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "whenever."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

gur

-0.72

âĵĺ

-0.70

FT

-0.69

hash

-0.69

roth

-0.68

Medic

-0.67

eg

-0.65

Met

-0.65

han

-0.64

asse

-0.64

POSITIVE LOGITS

soever

1.42

ĸļ

0.97

 occas

0.97

£ı

0.93

entimes

0.92

oun

0.90

 thereafter

0.88

ĪĴ

0.88

etheless

0.86

*/(

0.85

Activations Density 0.005%

the word "whenever" followed by an action or event

occurrences of the word "whenever."

No Comments

No Known Activations