INDEX

Explanations

date-related information such as years or time periods

oai_token-act-pair · gpt-3.5-turbo

sentences or phrases ending with a period

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 stamped

-0.80

 optional

-0.76

 trap

-0.70

 millisec

-0.70

 ration

-0.68

 punch

-0.67

 unpop

-0.65

 meaningless

-0.64

 explan

-0.64

 unavoidable

-0.62

POSITIVE LOGITS

 Additionally

1.24

 Currently

1.23

 Previously

1.22

 During

1.21

His

1.19

 Along

1.17

 Known

1.12

 Afterwards

1.12

He

1.11

 Serving

1.10

Activations Density 0.324%

date-related information such as years or time periods

sentences or phrases ending with a period

No Comments

No Known Activations