INDEX

Explanations

references to past associations or titles

oai_token-act-pair · gpt-3.5-turbo

references to individuals who hold or have held a significant role or position

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

antics

-0.80

hare

-0.78

\\\\\\\\

-0.76

ammy

-0.75

ourced

-0.73

atown

-0.70

otle

-0.70

urai

-0.70

mand

-0.69

Ped

-0.67

POSITIVE LOGITS

 Yugoslav

0.99

 Yugoslavia

0.89

est

0.78

 dictator

0.76

 Soviet

0.69

 owner

0.69

mentioned

0.68

 communist

0.68

 president

0.67

 convict

0.66

Activations Density 0.031%

references to past associations or titles

references to individuals who hold or have held a significant role or position

No Comments

No Known Activations