INDEX

Explanations

mentions of behaviors or attitudes of being complacent

oai_token-act-pair · gpt-3.5-turbo

terms related to cunning or deceptive behavior

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ilitating

-0.76

inges

-0.76

izophren

-0.70

oooooooooooooooo

-0.67

igslist

-0.65

iking

-0.65

ELF

-0.65

enez

-0.63

essional

-0.63

 Alam

-0.63

POSITIVE LOGITS

wolf

0.91

tank

0.86

rod

0.85

esis

0.83

cair

0.79

cel

0.78

ote

0.77

catentry

0.77

otes

0.76

osite

0.76

Activations Density 0.012%

mentions of behaviors or attitudes of being complacent

terms related to cunning or deceptive behavior

No Comments

No Known Activations