INDEX

Explanations

phrases indicating something is available at no cost

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the absence or removal of something

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

culosis

-0.90

pillar

-0.88

çīĪ

-0.77

apego

-0.76

phabet

-0.75

gars

-0.74

want

-0.73

abin

-0.73

along

-0.73

dad

-0.72

POSITIVE LOGITS

 mercy

0.78

 Sagan

0.71

any

0.69

 princip

0.68

 dominion

0.67

 perm

0.66

 Territory

0.66

 admitting

0.65

 mortal

0.65

 foreseeable

0.65

Activations Density 0.063%

phrases indicating something is available at no cost

phrases indicating the absence or removal of something

No Comments

No Known Activations