INDEX

Explanations

instances where actions or events are expressly denied or refuted

oai_token-act-pair · gpt-3.5-turbo

references to health issues or medical conditions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 humour

-0.77

 favourite

-0.74

 emot

-0.73

 witty

-0.71

 charms

-0.70

 appreciated

-0.69

 urges

-0.69

 colourful

-0.68

aph

-0.66

 favourites

-0.66

POSITIVE LOGITS

numbered

0.88

irteen

0.87

none

0.78

undred

0.77

REDACTED

0.74

None

0.72

ixty

0.71

ERE

0.70

total

0.70

each

0.69

Activations Density 1.155%

instances where actions or events are expressly denied or refuted

references to health issues or medical conditions

No Comments

No Known Activations