INDEX

Explanations

movie titles and TV show names, particularly those with a colon in the title

oai_token-act-pair · gpt-3.5-turbo

colons followed by titles or names of media properties

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

etooth

-0.87

uggest

-0.83

 tremend

-0.79

jri

-0.78

ebin

-0.77

ewitness

-0.75

inarily

-0.75

 fres

-0.74

gart

-0.72

 heated

-0.72

POSITIVE LOGITS

 Origins

0.87

 Legacy

0.80

 Reborn

0.80

 Chains

0.78

 Dangerous

0.77

 Ultimate

0.75

 Extended

0.75

 Imaging

0.75

 Portable

0.74

 Characters

0.74

Activations Density 0.045%

movie titles and TV show names, particularly those with a colon in the title

colons followed by titles or names of media properties

No Comments

No Known Activations