INDEX

Explanations

notifications or prompts of reassurance or advice in different contexts, emphasizing not to worry

oai_token-act-pair · gpt-3.5-turbo

reassurances related to not worrying

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

arb

-0.89

ewitness

-0.85

avorite

-0.81

pled

-0.73

artifacts

-0.72

arth

-0.71

 livest

-0.71

ĻĤ

-0.71

aw

-0.69

ingers

-0.68

POSITIVE LOGITS

warts

1.02

wart

0.90

 worry

0.87

lessly

0.87

 fret

0.87

ingly

0.82

 worrying

0.77

 ABOUT

0.77

 Lieberman

0.74

crow

0.72

Activations Density 0.020%

notifications or prompts of reassurance or advice in different contexts, emphasizing not to worry

reassurances related to not worrying

No Comments

No Known Activations