INDEX

Explanations

references to the TV network HBO

oai_token-act-pair · gpt-3.5-turbo

mentions of HBO and its programs or related content

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

acle

-0.83

acular

-0.83

uation

-0.81

omaly

-0.76

cy

-0.72

secut

-0.72

formation

-0.70

geant

-0.70

force

-0.70

leading

-0.69

POSITIVE LOGITS

HBO

0.87

EW

0.83

 Films

0.82

Mub

0.79

 premiered

0.77

ÃįÃį

0.76

 Networks

0.74

 Showtime

0.72

RECT

0.72

 Wire

0.71

Activations Density 0.007%

references to the TV network HBO

mentions of HBO and its programs or related content

No Comments

No Known Activations