INDEX

Explanations

words ending in '-wise' to compare different aspects or situations

oai_token-act-pair · gpt-3.5-turbo

references to wisdom or being "wise"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Nou

-0.76

IENCE

-0.65

Klu

-0.63

 Gazette

-0.60

 Shards

-0.59

 MÃ©

-0.58

 Acad

-0.57

 cellul

-0.56

 Chapters

-0.56

 Guantanamo

-0.56

POSITIVE LOGITS

wise

1.46

lihood

1.19

etheless

0.94

eyed

0.90

ly

0.87

liness

0.81

forward

0.77

nown

0.76

ably

0.76

folios

0.72

Activations Density 0.005%

words ending in '-wise' to compare different aspects or situations

references to wisdom or being "wise"

No Comments

No Known Activations