INDEX

Explanations

phrases related to denials or refusals

oai_token-act-pair · gpt-3.5-turbo

phrases indicating degrees of influence or impact

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

 Knot

-0.74

 Ashton

-0.72

incinn

-0.67

 proverb

-0.65

 staples

-0.63

 eyebrows

-0.63

 Corpus

-0.62

 Cumber

-0.62

 tales

-0.60

ŃĶ

-0.60

POSITIVE LOGITS

 whatsoever

1.15

 imaginable

0.90

soever

0.80

 resembling

0.78

 besides

0.77

isons

0.75

ames

0.74

brates

0.71

related

0.69

 resembles

0.69

Activations Density 0.027%

phrases related to denials or refusals

phrases indicating degrees of influence or impact

No Comments

No Known Activations