INDEX

Explanations

questions or commands directed towards the reader

oai_token-act-pair · gpt-3.5-turbo

interrogative words and phrases commonly associated with questioning and self-reflection

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

ocobo

-0.79

 Mehran

-0.72

EngineDebug

-0.69

 ÂŃ

-0.67

ICO

-0.66

javascript

-0.65

\":

-0.64

Ãª

-0.63

ATES

-0.62

SAN

-0.61

POSITIVE LOGITS

entimes

0.87

cknowled

0.82

cknow

0.78

 importantly

0.75

 blat

0.75

 honesty

0.73

eday

0.72

oret

0.72

neath

0.71

 furthermore

0.70

Activations Density 0.712%

questions or commands directed towards the reader

interrogative words and phrases commonly associated with questioning and self-reflection

No Comments

No Known Activations