INDEX

Explanations

words related to political or religious conflicts, particularly sectarianism

oai_token-act-pair · gpt-3.5-turbo

references to sectarian conflict

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

OTA

-0.82

nesota

-0.81

podcast

-0.80

OPA

-0.79

 Kinnikuman

-0.77

rey

-0.76

 Salon

-0.75

boards

-0.73

eor

-0.72

 Mercury

-0.71

POSITIVE LOGITS

 sectarian

1.01

 affili

0.98

 militias

0.93

 affiliation

0.92

 strife

0.91

 warfare

0.87

 divides

0.86

ities

0.83

 relations

0.82

 slurs

0.82

Activations Density 0.009%

words related to political or religious conflicts, particularly sectarianism

references to sectarian conflict

No Comments

No Known Activations