INDEX

Explanations

references to the word "pop"

oai_token-act-pair · gpt-3.5-turbo

references to pop culture

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

Contract

-0.90

 ASIC

-0.76

NEC

-0.71

 Investigator

-0.71

ridor

-0.69

ih

-0.68

oglu

-0.68

 Fortress

-0.67

cedented

-0.67

ieves

-0.66

POSITIVE LOGITS

pop

3.75

pop

2.29

 pops

2.28

Pop

2.25

 popping

1.96

 popup

1.89

 popped

1.88

Pop

1.85

POP

1.70

 populate

1.19

Activations Density 0.013%

references to the word "pop"

references to pop culture

No Comments

No Known Activations