INDEX

Explanations

phrases related to undercover operations or investigations

oai_token-act-pair · gpt-3.5-turbo

references to undercover activities or operations

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

itsch

-0.76

gran

-0.75

Pwr

-0.73

È

-0.71

ogun

-0.70

 congr

-0.70

grave

-0.69

cing

-0.67

RR

-0.66

hetti

-0.65

POSITIVE LOGITS

 informant

1.08

 undercover

1.07

 informants

0.89

 infiltr

0.83

 sting

0.80

 mole

0.79

 plotting

0.76

 spying

0.75

 detective

0.75

 operatives

0.74

Activations Density 0.026%

phrases related to undercover operations or investigations

references to undercover activities or operations

No Comments

No Known Activations