INDEX

Explanations

text related to cities, regions, and political events

oai_token-act-pair · gpt-3.5-turbo

specific geographical locations and associated political regions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 bandwagon

-0.73

STD

-0.70

Reviewer

-0.65

 Catalyst

-0.63

adders

-0.62

 Architects

-0.62

eus

-0.60

 Collider

-0.60

 Trayvon

-0.58

tle

-0.57

POSITIVE LOGITS

 northeastern

1.33

 southwestern

1.32

 northwestern

1.31

 southeastern

1.28

 province

1.22

 northern

1.18

 northeast

1.17

 southwest

1.17

 northwest

1.16

 southern

1.14

Activations Density 0.376%

text related to cities, regions, and political events

specific geographical locations and associated political regions

No Comments

No Known Activations