INDEX

Explanations

formatting characters, particularly when used in code

oai_token-act-pair · gpt-3.5-turbo

patterns related to programming and coding syntax

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

 paran

-0.86

ĸļ

-0.78

 Vill

-0.75

 Leth

-0.72

 Manit

-0.71

 mathemat

-0.68

 Kath

-0.68

Sau

-0.68

 travellers

-0.66

 Stew

-0.66

POSITIVE LOGITS

ERROR

1.07

};

1.06

/*

1.05

Invalid

1.04

//

0.99

@@

0.98

assert

0.93

Http

0.93

&&

0.92

DEBUG

0.90

Activations Density 0.094%

formatting characters, particularly when used in code

patterns related to programming and coding syntax

No Comments

No Known Activations