INDEX

Explanations

phrases related to confessing or admissions of truth

oai_token-act-pair · gpt-3.5-turbo

terms related to admissions and confessions of wrongdoing

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 upkeep

-0.67

 braces

-0.64

 Orchestra

-0.64

isites

-0.63

scill

-0.62

 odds

-0.61

PDATE

-0.61

axy

-0.61

chair

-0.60

 queues

-0.60

POSITIVE LOGITS

 confess

1.01

 confessions

0.99

 confession

0.95

itives

0.94

itive

0.81

 confessed

0.80

sov

0.78

essional

0.74

cia

0.73

 incrim

0.72

Activations Density 0.040%

phrases related to confessing or admissions of truth

terms related to admissions and confessions of wrongdoing

No Comments

No Known Activations