INDEX

Explanations

mentions of the term 'Murd*er' with different variations, along with some related terms like 'Daredevil'

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "Murderer" and references to the show "Daredevil."

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

 conditioning

-0.71

 placement

-0.71

 Gemini

-0.67

 corrective

-0.65

 magnification

-0.64

zed

-0.62

condition

-0.62

 selective

-0.62

eq

-0.62

graded

-0.62

POSITIVE LOGITS

 Murd

1.37

stead

0.91

erer

0.90

erers

0.89

erest

0.81

aby

0.80

aredevil

0.78

anth

0.77

luaj

0.77

ings

0.76

Activations Density 0.013%

mentions of the term 'Murd*er' with different variations, along with some related terms like 'Daredevil'

occurrences of the word "Murderer" and references to the show "Daredevil."

No Comments

No Known Activations