INDEX

Explanations

mentions of specific actions or events happening in a sequence

oai_token-act-pair · gpt-3.5-turbo

phrases related to legal proceedings and crime

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

Kansas

-0.93

Newsletter

-0.92

 favored

-0.86

Washington

-0.85

Zip

-0.79

ospons

-0.78

Chicago

-0.76

 favor

-0.76

Iowa

-0.73

 organizational

-0.73

POSITIVE LOGITS

 realise

1.20

 apologised

1.16

 organisers

1.16

 apologise

1.16

 paed

1.13

 footballer

1.10

 recognise

1.07

 criticised

1.06

mum

1.05

 analyse

1.04

Activations Density 1.248%

mentions of specific actions or events happening in a sequence

phrases related to legal proceedings and crime

No Comments

No Known Activations