INDEX

Explanations

large numerical quantities, specifically the word "million."

oai_token-act-pair · gpt-3.5-turbo

references to large quantities or populations, specifically "million" or numbers close to it

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

anwhile

-0.81

sonian

-0.79

comes

-0.75

abulary

-0.75

agon

-0.73

strength

-0.72

date

-0.70

dates

-0.69

emia

-0.69

dated

-0.69

POSITIVE LOGITS

 dollars

0.94

aire

0.86

aires

0.85

 crore

0.82

+)

0.78

 hectares

0.78

iles

0.75

apixel

0.75

 euros

0.75

yen

0.75

Activations Density 0.043%

large numerical quantities, specifically the word "million."

references to large quantities or populations, specifically "million" or numbers close to it

No Comments

No Known Activations