INDEX

Explanations

references to puppetry and mascots

oai_token-act-pair · gpt-3.5-turbo

references to puppets and puppet-related media

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

neau

-0.84

yer

-0.82

imentary

-0.76

iam

-0.72

ikh

-0.67

 appointments

-0.66

prus

-0.64

ient

-0.63

ugal

-0.63

 embargo

-0.62

POSITIVE LOGITS

 mascot

1.16

 costumes

0.98

 Pikachu

0.96

 Simpsons

0.94

 masc

0.94

 Costume

0.93

pole

0.93

oji

0.90

pants

0.87

 plush

0.86

Activations Density 0.067%

references to puppetry and mascots

references to puppets and puppet-related media

No Comments

No Known Activations