INDEX

Explanations

averages and totals of numerical values in a range of contexts

oai_token-act-pair · gpt-3.5-turbo

terms related to averages and totals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 changed

-0.69

did

-0.69

 benefited

-0.68

 Changed

-0.68

atari

-0.67

 fled

-0.67

bek

-0.66

 depended

-0.66

 fought

-0.66

 acted

-0.65

POSITIVE LOGITS

 averaging

3.47

 totaling

1.47

 estimating

1.41

 weighing

1.36

 topping

1.33

 costing

1.27

 rounding

1.22

 measuring

1.21

 numbering

1.18

 scoring

1.15

Activations Density 0.022%

averages and totals of numerical values in a range of contexts

terms related to averages and totals

No Comments

No Known Activations