INDEX

Explanations

proper nouns, specifically the name "Anne", with varying degrees of emphasis indicated by the activation values

oai_token-act-pair · gpt-3.5-turbo

mentions of the name "Anne."

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

ierrez

-0.88

yz

-0.77

uador

-0.75

committee

-0.72

milo

-0.72

ornia

-0.72

raltar

-0.72

medium

-0.72

ython

-0.72

ivas

-0.71

POSITIVE LOGITS

 Anne

1.37

Anne

1.36

 Marie

1.03

Marie

0.96

 Joan

0.82

 Hath

0.82

alogy

0.80

 Helen

0.80

issance

0.78

 Margaret

0.77

Activations Density 0.006%

proper nouns, specifically the name "Anne", with varying degrees of emphasis indicated by the activation values

mentions of the name "Anne."

No Comments

No Known Activations