INDEX

Explanations

mentions of names or terms that have been changed from their previous identification

oai_token-act-pair · gpt-3.5-turbo

the term "formerly" followed by a reference to past identities or names of entities

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

eer

-0.91

pled

-0.87

anche

-0.86

oos

-0.77

eers

-0.76

otle

-0.75

eanor

-0.74

jriwal

-0.73

antics

-0.73

ggle

-0.72

POSITIVE LOGITS

 defunct

1.04

 obsolete

0.94

formerly

0.87

 known

0.85

 extinct

0.81

adays

0.79

 founded

0.79

 housed

0.78

 Known

0.77

lehem

0.76

Activations Density 0.034%

mentions of names or terms that have been changed from their previous identification

the term "formerly" followed by a reference to past identities or names of entities

No Comments

No Known Activations