INDEX

Explanations

phrases related to hiding or concealing

oai_token-act-pair · gpt-3.5-turbo

references to the concept of concealment and deception

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

 GEAR

-0.70

ucky

-0.69

 Articles

-0.67

fleet

-0.65

 Hits

-0.61

 Archdemon

-0.60

 conductor

-0.60

 Bucks

-0.60

 Rite

-0.58

 GOODMAN

-0.57

POSITIVE LOGITS

ivably

1.28

iving

1.07

veland

0.89

ences

0.88

rence

0.83

 conce

0.82

ivers

0.81

aling

0.81

uded

0.80

bes

0.79

Activations Density 0.012%

phrases related to hiding or concealing

references to the concept of concealment and deception

No Comments

No Known Activations