INDEX

Explanations

the number "two" in various contexts

oai_token-act-pair · gpt-3.5-turbo

numerical references to quantities or durations, particularly the words "two" and "three."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 embassies

-0.63

 Acad

-0.62

akings

-0.61

assies

-0.57

 Exper

-0.56

FORE

-0.55

 Indust

-0.54

soDeliveryDate

-0.54

 Boards

-0.52

 Nasa

-0.52

POSITIVE LOGITS

fold

1.31

1.19

 dimensional

1.13

peat

1.12

teenth

1.10

âĢĳ

1.05

some

1.02

âĢĲ

1.02

-,

1.00

 sided

0.97

Activations Density 0.074%

the number "two" in various contexts

numerical references to quantities or durations, particularly the words "two" and "three."

No Comments

No Known Activations