INDEX

Explanations

mentions of the word "Kind" followed by a contextually relevant word

oai_token-act-pair · gpt-3.5-turbo

references to kindness and related themes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Downloadha

-0.87

uters

-0.70

ptions

-0.70

OPE

-0.69

pez

-0.68

è¦ļéĨĴ

-0.67

ITNESS

-0.66

asts

-0.66

opers

-0.66

IVER

-0.65

POSITIVE LOGITS

 Kind

0.94

kind

0.92

ness

0.84

hearted

0.82

entimes

0.79

liness

0.76

iciary

0.75

lich

0.74

red

0.73

 Pair

0.72

Activations Density 0.017%

mentions of the word "Kind" followed by a contextually relevant word

references to kindness and related themes

No Comments

No Known Activations