INDEX

Explanations

numbers in a sequence, particularly related to rankings or orders

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "number" along with related numerical terms

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

rador

-0.82

 Shroud

-0.76

clair

-0.68

Interstitial

-0.68

axter

-0.67

Materials

-0.66

INTON

-0.64

aples

-0.64

ossier

-0.62

rained

-0.61

POSITIVE LOGITS

plates

0.95

number

0.84

plate

0.81

 number

0.74

 Numbers

0.73

enance

0.72

 numbers

0.72

metry

0.72

âĢİ

0.71

 digits

0.70

Activations Density 0.039%

numbers in a sequence, particularly related to rankings or orders

occurrences of the word "number" along with related numerical terms

No Comments

No Known Activations