INDEX

Explanations

instances of the prefix "ex-" used before a word

oai_token-act-pair · gpt-3.5-turbo

references to past relationships or entities marked as "ex" or former

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 shading

-0.83

uden

-0.77

 amplification

-0.74

 weighted

-0.73

 underlying

-0.73

 dances

-0.73

 elong

-0.72

 constrained

-0.71

 digest

-0.70

 imagination

-0.70

POSITIVE LOGITS

girlfriend

1.63

husband

1.49

employ

1.47

wife

1.43

military

1.42

football

1.33

serv

1.32

members

1.32

CIA

1.31

slave

1.29

Activations Density 0.017%

instances of the prefix "ex-" used before a word

references to past relationships or entities marked as "ex" or former

No Comments

No Known Activations