INDEX

Explanations

phrases related to defending or justifying actions or beliefs

oai_token-act-pair · gpt-3.5-turbo

instances of the word "defend" and its variations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

hall

-0.69

nown

-0.67

Ju

-0.66

outh

-0.64

Hop

-0.63

orrow

-0.63

mad

-0.63

 explode

-0.62

NetMessage

-0.62

RESULTS

-0.61

POSITIVE LOGITS

 defending

0.94

 against

0.89

 defends

0.81

atively

0.79

 defend

0.76

 Against

0.76

 defended

0.73

orate

0.72

unct

0.70

 Defense

0.68

Activations Density 0.020%

phrases related to defending or justifying actions or beliefs

instances of the word "defend" and its variations

No Comments

No Known Activations