INDEX

Explanations

phrases related to a specific person or name, in this case, "Watts"

oai_token-act-pair · gpt-3.5-turbo

instances of the name "Watts."

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

oral

-0.72

*/(

-0.70

onyms

-0.70

istically

-0.68

vice

-0.66

opsy

-0.64

viously

-0.64

amon

-0.63

OPLE

-0.62

 Lyft

-0.61

POSITIVE LOGITS

 Watts

1.44

atts

1.31

 Compton

0.85

cliffe

0.83

acre

0.82

leys

0.81

halla

0.79

burgh

0.77

 watt

0.76

 Gunn

0.75

Activations Density 0.005%

phrases related to a specific person or name, in this case, "Watts"

instances of the name "Watts."

No Comments

No Known Activations