INDEX

Explanations

keywords related to cunning or deceitful behavior

oai_token-act-pair · gpt-3.5-turbo

terms related to deception and malicious intentions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

Recomm

-0.71

Ê

-0.71

ains

-0.71

anan

-0.68

ainers

-0.67

--+

-0.67

ikini

-0.67

grain

-0.66

UTC

-0.66

birth

-0.66

POSITIVE LOGITS

 plotting

1.10

 intrig

1.04

 sche

0.98

 intrigue

0.97

 cunning

0.96

 mastermind

0.96

eering

0.96

 twist

0.94

 disgu

0.94

 plotted

0.94

Activations Density 0.099%

keywords related to cunning or deceitful behavior

terms related to deception and malicious intentions

No Comments

No Known Activations