INDEX

Explanations

locations or names ending in "ondo"

oai_token-act-pair · gpt-3.5-turbo

proper nouns and names

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Pont

-0.73

 corpor

-0.68

 Fine

-0.66

Cum

-0.66

 Russ

-0.62

Bed

-0.61

 Belt

-0.60

 dice

-0.60

 Leth

-0.60

 Brennan

-0.59

POSITIVE LOGITS

ako

4.05

uko

1.39

emo

1.36

oko

1.27

amaru

1.23

oka

1.23

uku

1.18

aka

1.18

aki

1.15

allo

1.14

Activations Density 0.010%

locations or names ending in "ondo"

proper nouns and names

No Comments

No Known Activations