INDEX

Explanations

the word "rebel" and related terms in the text

oai_token-act-pair · gpt-3.5-turbo

references to rebel groups or factions

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

ographies

-0.88

thora

-0.83

è¯

-0.75

 Robo

-0.74

ocobo

-0.74

gow

-0.74

 Kramer

-0.74

othy

-0.73

 Hitchcock

-0.73

icrobial

-0.72

POSITIVE LOGITS

 factions

1.01

 milit

1.00

 rebels

0.96

 rebel

0.93

 faction

0.92

 shelling

0.89

 army

0.87

 militias

0.85

 camps

0.85

 waging

0.85

Activations Density 0.025%

the word "rebel" and related terms in the text

references to rebel groups or factions

No Comments

No Known Activations