INDEX

Explanations

references to powerful and intimidating entities or creatures

oai_token-act-pair · gpt-3.5-turbo

references to 'beasts' or animal-like metaphors

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

pai

-0.83

arters

-0.79

encing

-0.79

psons

-0.76

bered

-0.75

enza

-0.75

monds

-0.73

licted

-0.71

endiary

-0.70

earcher

-0.67

POSITIVE LOGITS

 beasts

1.04

 Beasts

0.95

 beast

0.89

 carc

0.88

mong

0.85

lords

0.82

bags

0.80

 Beast

0.80

zilla

0.78

ishly

0.77

Activations Density 0.020%

references to powerful and intimidating entities or creatures

references to 'beasts' or animal-like metaphors

No Comments

No Known Activations