INDEX

Explanations

the word "people" occurring multiple times in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to "people" and their roles or experiences

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

BALL

-0.79

FIN

-0.75

forward

-0.70

Va

-0.69

ANA

-0.67

POSE

-0.65

CONT

-0.65

inventoryQuantity

-0.65

REM

-0.65

LV

-0.65

POSITIVE LOGITS

folk

1.00

 inhab

0.91

 smugglers

0.90

liest

0.84

 sacrificed

0.79

 taxp

0.78

 entrusted

0.77

ourced

0.76

 closest

0.75

ongs

0.75

Activations Density 0.057%

the word "people" occurring multiple times in various contexts

references to "people" and their roles or experiences

No Comments

No Known Activations