INDEX

Explanations

names, possibly of people

oai_token-act-pair · gpt-3.5-turbo

proper nouns, particularly names associated with specific individuals or entities

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

 Wonderland

-0.64

"{

-0.62

 ãĢĮ

-0.58

 Blacks

-0.57

 accuser

-0.57

 Parables

-0.57

ylum

-0.57

 hereafter

-0.56

othing

-0.56

 Kinnikuman

-0.56

POSITIVE LOGITS

ï¸ı

0.79

ooth

0.77

twitch

0.73

cone

0.72

 cooker

0.72

cker

0.72

oise

0.71

cular

0.69

icky

0.69

PDATE

0.66

Activations Density 0.212%

names, possibly of people

proper nouns, particularly names associated with specific individuals or entities

No Comments

No Known Activations