INDEX

Explanations

proper nouns related to personal entities

oai_token-act-pair · gpt-3.5-turbo

mentions of "bans" and "Bansal."

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

Ocean

-0.79

rious

-0.77

Union

-0.74

selves

-0.73

 Lakes

-0.70

Sea

-0.70

×Ķ

-0.69

 IMAGES

-0.68

Dug

-0.68

Cor

-0.68

POSITIVE LOGITS

hee

1.25

 bans

1.24

 Bans

1.11

wana

0.94

ourgeois

0.91

creen

0.89

uits

0.89

ban

0.88

 outlaw

0.86

heon

0.85

Activations Density 0.009%

proper nouns related to personal entities

mentions of "bans" and "Bansal."

No Comments

No Known Activations