INDEX

Explanations

highly emphasized and distinct terms within a text

oai_token-act-pair · gpt-3.5-turbo

instances of the term "int" or related variations indicating intensity, interaction, or introspection

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Accessory

-0.68

hyde

-0.67

wagen

-0.62

bial

-0.59

awaru

-0.59

 Mbps

-0.58

 condos

-0.57

Fram

-0.57

ggles

-0.56

 innovation

-0.55

POSITIVE LOGITS

itial

1.10

ensions

0.95

ract

0.87

itially

0.86

ruct

0.86

ension

0.86

repre

0.84

ents

0.82

rep

0.80

ently

0.79

Activations Density 0.016%

highly emphasized and distinct terms within a text

instances of the term "int" or related variations indicating intensity, interaction, or introspection

No Comments

No Known Activations