INDEX

Explanations

proper nouns that consist of two short words

oai_token-act-pair · gpt-3.5-turbo

references to "Peep Show," indicating a focus on content related to this specific television show

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

schild

-1.00

 EDITION

-0.78

stretched

-0.74

etheless

-0.73

orage

-0.71

 Balanced

-0.70

 GOODMAN

-0.70

-0.69

EngineDebug

-0.68

naire

-0.67

POSITIVE LOGITS

anut

1.33

anuts

1.12

asant

1.10

eps

1.07

ek

1.02

ggie

0.97

eking

0.97

aky

0.97

pperc

0.95

ep

0.94

Activations Density 0.017%

proper nouns that consist of two short words

references to "Peep Show," indicating a focus on content related to this specific television show

No Comments

No Known Activations