INDEX

Explanations

titles or mentions of war-related content

oai_token-act-pair · gpt-3.5-turbo

references to the concept of war

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

sembly

-0.90

essee

-0.88

Ħ¢

-0.85

aunder

-0.79

ħĭ

-0.79

issance

-0.77

İĭ

-0.72

 incorpor

-0.72

aminer

-0.70

 livest

-0.67

POSITIVE LOGITS

rior

1.30

riors

1.23

fare

1.22

lord

1.19

lords

1.19

locks

1.07

restling

0.98

ping

0.94

zone

0.93

ped

0.93

Activations Density 0.028%

titles or mentions of war-related content

references to the concept of war

No Comments

No Known Activations