INDEX

Explanations

the word "broken" with varying degrees of severity

oai_token-act-pair · gpt-3.5-turbo

phrases that include the word "broken."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

minist

-0.79

utterstock

-0.75

azor

-0.75

yss

-0.75

erva

-0.75

metics

-0.74

ickr

-0.73

idency

-0.73

itatively

-0.72

Reviewer

-0.72

POSITIVE LOGITS

neck

0.89

 bones

0.86

 broken

0.85

bones

0.79

broken

0.78

 necks

0.75

 fracture

0.73

 Broken

0.71

 broke

0.70

breaks

0.70

Activations Density 0.016%

the word "broken" with varying degrees of severity

phrases that include the word "broken."

No Comments

No Known Activations