INDEX

Explanations

phrases related to different episodes or parts of a series

oai_token-act-pair · gpt-3.5-turbo

references to specific episodes in a series

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

FUL

-0.92

LESS

-0.82

ãĥīãĥ©ãĤ´ãĥ³

-0.73

IFIC

-0.71

WAYS

-0.70

Rid

-0.66

 Pathfinder

-0.66

^^^^

-0.66

hips

-0.66

CLASSIFIED

-0.65

POSITIVE LOGITS

iphany

1.25

isodes

1.25

hemer

1.15

istle

1.13

iscopal

1.06

igen

0.99

och

0.93

idem

0.92

oton

0.92

iph

0.90

Activations Density 0.009%

phrases related to different episodes or parts of a series

references to specific episodes in a series

No Comments

No Known Activations