INDEX

Explanations

phrases related to exoneration or justification

oai_token-act-pair · gpt-3.5-turbo

terms related to legal concepts and sensationalism in reporting

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

erker

-0.81

eem

-0.78

iem

-0.74

ublic

-0.73

assian

-0.71

eworld

-0.70

endor

-0.69

ropy

-0.69

sterdam

-0.69

atcher

-0.68

POSITIVE LOGITS

IFIED

0.82

LV

0.78

0.76

0.74

 "$:/

0.73

ized

0.70

wcsstore

0.69

idated

0.68

agate

0.66

ruciating

0.66

Activations Density 0.026%

phrases related to exoneration or justification

terms related to legal concepts and sensationalism in reporting

No Comments

No Known Activations