INDEX

Explanations

the word "boys" with varying activations, emphasizing a focus on this specific term

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "boys."

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

mediated

-0.73

uncture

-0.71

Accessory

-0.69

Sharp

-0.69

aeda

-0.68

ãĥķãĤ©

-0.67

osi

-0.66

ascular

-0.66

inventoryQuantity

-0.66

itures

-0.66

POSITIVE LOGITS

 boys

0.97

 Scouts

0.97

friend

0.85

hood

0.83

boys

0.81

hift

0.79

 puberty

0.78

 ages

0.77

 Boys

0.77

 girls

0.76

Activations Density 0.013%

the word "boys" with varying activations, emphasizing a focus on this specific term

occurrences of the word "boys."

No Comments

No Known Activations