INDEX

Explanations

phrases with the word "two"

oai_token-act-pair · gpt-3.5-turbo

references to the number two

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ugu

-0.77

asta

-0.75

atown

-0.72

iggins

-0.68

annel

-0.68

ubs

-0.68

ushima

-0.67

yz

-0.66

renheit

-0.66

ovi

-0.66

POSITIVE LOGITS

 halves

1.23

 thirds

1.19

fold

1.02

 sexes

0.95

 sides

0.89

 dozen

0.88

teenth

0.85

 Kore

0.84

 main

0.80

 hundred

0.79

Activations Density 0.054%

phrases with the word "two"

references to the number two

No Comments

No Known Activations