INDEX

Explanations

phrases containing the word "attacked"

oai_token-act-pair · gpt-3.5-turbo

instances of the word "attacked."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Vert

-0.72

val

-0.65

YC

-0.63

tz

-0.63

shown

-0.61

atom

-0.61

sa

-0.61

 vert

-0.61

flu

-0.61

aver

-0.61

POSITIVE LOGITS

attack

1.03

 attacked

0.94

 attacks

0.94

 attackers

0.89

oise

0.89

 attack

0.87

ritch

0.86

 attacking

0.85

ivated

0.82

Attack

0.80

Activations Density 0.015%

phrases containing the word "attacked"

instances of the word "attacked."

No Comments

No Known Activations