INDEX

Explanations

words related to legal consent or authorization

oai_token-act-pair · gpt-3.5-turbo

references to consent in various contexts

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Tycoon

-0.92

Lens

-0.74

Closure

-0.72

Dak

-0.67

 Feld

-0.64

 Hebdo

-0.63

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

-0.61

ches

-0.60

Spot

-0.60

ched

-0.60

POSITIVE LOGITS

 consent

1.04

 Consent

0.94

 decree

0.91

ajor

0.89

 decre

0.78

itus

0.77

 authorizing

0.77

uary

0.75

ibility

0.74

itimate

0.74

Activations Density 0.014%

words related to legal consent or authorization

references to consent in various contexts

No Comments

No Known Activations