INDEX

Explanations

phrases indicating beliefs or faith in something

oai_token-act-pair · gpt-3.5-turbo

expressions of belief

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

soever

-1.00

DragonMagazine

-0.79

ener

-0.70

MAP

-0.69

stroke

-0.68

IRO

-0.66

edin

-0.66

hai

-0.65

enegger

-0.65

filename

-0.63

POSITIVE LOGITS

clus

1.13

 miracles

1.09

 spite

0.87

 preserving

0.85

 fairness

0.82

 miraculous

0.81

 moderation

0.80

effic

0.79

 independence

0.78

 equality

0.77

Activations Density 0.058%

phrases indicating beliefs or faith in something

expressions of belief

No Comments

No Known Activations