INDEX

Explanations

phrases that suggest conspiracy theories or coordinated efforts to manipulate events

oai_token-act-pair · gpt-3.5-turbo

phrases indicating intentional actions or plots

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

answered

-0.82

TPPStreamerBot

-0.79

felt

-0.79

accompan

-0.75

aton

-0.73

Bei

-0.72

apache

-0.72

 Jere

-0.71

living

-0.70

checked

-0.70

POSITIVE LOGITS

 intimidate

1.57

 undermine

1.56

 deceive

1.54

 manipulate

1.48

 discredit

1.47

 mislead

1.41

 exploit

1.37

 confuse

1.37

 deprive

1.37

 blackmail

1.36

Activations Density 0.269%

phrases that suggest conspiracy theories or coordinated efforts to manipulate events

phrases indicating intentional actions or plots

No Comments

No Known Activations