INDEX

Explanations

contractions where 'n is followed by t

oai_token-act-pair · gpt-3.5-turbo

the phrase "didn't" indicating a sense of regret or inability to do something

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

anwhile

-0.88

 eleph

-0.82

 neighb

-0.76

 populated

-0.75

 ingred

-0.75

 metic

-0.70

 princ

-0.70

everal

-0.69

 compan

-0.69

 gobl

-0.69

POSITIVE LOGITS

't

1.80

ÃŃ

1.06

Â´

0.99

ned

0.97

itely

0.90

nt

0.87

gered

0.86

ald

0.83

Ts

0.83

Ì¶

0.82

Activations Density 0.094%

contractions where 'n is followed by t

the phrase "didn't" indicating a sense of regret or inability to do something

No Comments

No Known Activations