INDEX

Explanations

instances of the word "alleged" in a text

oai_token-act-pair · gpt-3.5-turbo

references to alleged incidents or claims of wrongdoing

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

 guiActiveUnfocused

-0.73

 Clicker

-0.71

ARCH

-0.70

pb

-0.69

eyes

-0.68

OVA

-0.66

bern

-0.65

ajo

-0.63

beard

-0.63

patch

-0.63

POSITIVE LOGITS

iary

0.87

 allegations

0.85

ities

0.85

 accusations

0.79

 abuser

0.78

 accuser

0.77

edly

0.74

 misrepresent

0.74

 violations

0.72

 abuse

0.72

Activations Density 0.025%

instances of the word "alleged" in a text

references to alleged incidents or claims of wrongdoing

No Comments

No Known Activations