INDEX

Explanations

words related to high-ranking positions or entities

oai_token-act-pair · gpt-3.5-turbo

references to levels of intensity or importance

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 Lucia

-0.80

 Andromeda

-0.77

Frames

-0.77

OPLE

-0.76

 Predators

-0.72

Meg

-0.69

Woman

-0.67

 Stars

-0.67

ELY

-0.67

iframe

-0.66

POSITIVE LOGITS

level

0.91

 levels

0.83

 level

0.82

acies

0.71

 diagrams

0.69

 guidance

0.68

ahead

0.67

rise

0.67

 Level

0.65

tenance

0.64

Activations Density 0.015%

words related to high-ranking positions or entities

references to levels of intensity or importance

No Comments

No Known Activations