INDEX

Explanations

proper nouns or names of people and organizations related to news or reports

oai_token-act-pair · gpt-3.5-turbo

proper nouns, particularly names and titles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

Defense

-0.79

 nonexistent

-0.79

ADVERTISEMENT

-0.78

 neighboring

-0.75

 caliber

-0.74

 harbor

-0.74

 demeanor

-0.74

adena

-0.73

imize

-0.73

 armored

-0.72

POSITIVE LOGITS

 apologised

1.75

 criticised

1.56

 recognised

1.28

 realised

1.28

 emphas

1.25

QC

1.24

 analysed

1.21

 learnt

1.20

 said

1.19

 insisted

1.15

Activations Density 0.370%

proper nouns or names of people and organizations related to news or reports

proper nouns, particularly names and titles

No Comments

No Known Activations