INDEX

Explanations

phrases indicating conflict or contradiction

oai_token-act-pair · gpt-3.5-turbo

references to policies and contradictions in ideological positions

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

naissance

-0.69

culosis

-0.65

 Swordsman

-0.63

icides

-0.62

 Assist

-0.62

letes

-0.61

replace

-0.61

assisted

-0.61

iard

-0.61

missing

-0.61

POSITIVE LOGITS

 ideals

1.40

 orthodoxy

1.39

 tenets

1.37

 ethos

1.36

 principles

1.35

 teachings

1.31

 norms

1.30

 beliefs

1.30

 worldview

1.26

 philosophies

1.25

Activations Density 0.301%

phrases indicating conflict or contradiction

references to policies and contradictions in ideological positions

No Comments

No Known Activations