INDEX

Explanations

numbers or numerical patterns

oai_token-act-pair · gpt-3.5-turbo

mentions of numerical data and statistics

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

clair

-0.74

idden

-0.71

rift

-0.64

merce

-0.63

perty

-0.63

pired

-0.62

phrine

-0.61

rend

-0.61

ovie

-0.60

isure

-0.60

POSITIVE LOGITS

 Number

1.18

 Numbers

1.10

Number

1.07

number

0.96

NUM

0.95

 number

0.89

 numbers

0.89

NUM

0.88

=#

0.80

 Deaths

0.79

Activations Density 0.015%

numbers or numerical patterns

mentions of numerical data and statistics

No Comments

No Known Activations