INDEX

Explanations

references to product-related terms, possibly related to marketing or development

oai_token-act-pair · gpt-3.5-turbo

references to products and their features

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

laus

-0.72

 Brother

-0.67

apo

-0.64

 Shades

-0.64

CLASSIFIED

-0.63

arning

-0.63

aft

-0.62

 Manor

-0.61

 Hurricanes

-0.60

 millenn

-0.60

POSITIVE LOGITS

ivity

1.19

ively

1.06

ivities

0.86

Interstitial

0.84

iveness

0.83

 packaging

0.82

Products

0.80

Hunt

0.79

kit

0.74

icons

0.72

Activations Density 0.022%

references to product-related terms, possibly related to marketing or development

references to products and their features

No Comments

No Known Activations