INDEX

Explanations

the word "person"

oai_token-act-pair · gpt-3.5-turbo

references to individuals or persons in various contexts

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

enthal

-0.74

 actionGroup

-0.73

eryl

-0.66

Tx

-0.65

 Equal

-0.65

onder

-0.64

word

-0.62

DL

-0.62

ä½¿

-0.62

 Triangle

-0.61

POSITIVE LOGITS

hood

1.07

nel

0.82

wise

0.74

 vaccinated

0.71

riage

0.71

 person

0.70

WithNo

0.70

mone

0.69

 somet

0.69

ifest

0.69

Activations Density 0.025%

the word "person"

references to individuals or persons in various contexts

No Comments

No Known Activations