INDEX

Explanations

terms related to cooperation, coordination, conspiracy, or illegal collaboration

oai_token-act-pair · gpt-3.5-turbo

terms related to collusion and complicity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

=-=-=-=-=-=-=-=-

-0.75

thur

-0.72

ĻĤ

-0.69

ulton

-0.69

alg

-0.68

Hop

-0.67

mac

-0.67

fly

-0.67

 Balanced

-0.66

birth

-0.65

POSITIVE LOGITS

 collusion

1.22

 conspiring

0.88

 tampering

0.84

 complicit

0.82

 complicity

0.80

 unfocusedRange

0.73

 linking

0.71

 conspir

0.71

 wink

0.71

 between

0.71

Activations Density 0.011%

terms related to cooperation, coordination, conspiracy, or illegal collaboration

terms related to collusion and complicity

No Comments

No Known Activations