INDEX

Explanations

references to passengers or passenger-related information

oai_token-act-pair · gpt-3.5-turbo

references to passengers in various contexts

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

arios

-0.96

laus

-0.83

 Skydragon

-0.80

endo

-0.80

xual

-0.79

enda

-0.77

earchers

-0.76

krit

-0.76

cale

-0.75

sembly

-0.75

POSITIVE LOGITS

 passenger

1.01

wings

0.89

 compartment

0.88

 pige

0.85

 Seat

0.82

 pigeon

0.81

 cabin

0.79

 cars

0.76

 airliner

0.76

 airline

0.76

Activations Density 0.010%

references to passengers or passenger-related information

references to passengers in various contexts

No Comments

No Known Activations