INDEX

Explanations

adjectives describing a level or intensity of something

oai_token-act-pair · gpt-3.5-turbo

phrases that highlight absurdity or contradictions

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

iller

-0.73

çīĪ

-0.67

FTWARE

-0.67

NL

-0.66

ashington

-0.64

Published

-0.63

ilaterally

-0.62

cellaneous

-0.60

 showc

-0.60

 Mandatory

-0.59

POSITIVE LOGITS

lihood

0.77

 hell

0.75

 peas

0.73

iah

0.71

daq

0.70

par

0.70

 apples

0.66

heed

0.66

phy

0.65

apple

0.64

Activations Density 0.108%

adjectives describing a level or intensity of something

phrases that highlight absurdity or contradictions

No Comments

No Known Activations