INDEX

Explanations

locations and incidents related to conflict and violence

oai_token-act-pair · gpt-3.5-turbo

references to specific geographic locations and associated events

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

Knight

-0.77

NSA

-0.76

Reviewer

-0.75

 mentors

-0.74

-0.73

CLASSIFIED

-0.71

STEM

-0.71

 Lifetime

-0.71

 Founders

-0.70

Privacy

-0.70

POSITIVE LOGITS

 countryside

1.14

 suburb

1.10

 suburbs

1.08

 outskirts

1.08

 ravaged

1.07

 southeastern

1.01

 devastated

1.01

 neighbourhoods

1.01

 mountainous

1.01

 eastern

0.99

Activations Density 0.748%

locations and incidents related to conflict and violence

references to specific geographic locations and associated events

No Comments

No Known Activations