INDEX

Explanations

phrases related to deception or misleading actions

oai_token-act-pair · gpt-3.5-turbo

references to individuals or groups being misled, scared, or shocked in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 enhancement

-0.62

ITNESS

-0.60

 flourishing

-0.57

 fitt

-0.57

OPA

-0.56

 Replacement

-0.55

 redistribution

-0.55

 accomp

-0.54

 illumination

-0.54

 automation

-0.54

POSITIVE LOGITS

 unnecessarily

0.82

 into

0.81

selves

0.81

 merciless

0.79

 senseless

0.75

 gull

0.71

ores

0.69

into

0.68

 Into

0.67

 sensibilities

0.67

Activations Density 0.292%

phrases related to deception or misleading actions

references to individuals or groups being misled, scared, or shocked in various contexts

No Comments

No Known Activations