INDEX

Explanations

names related to people or entities, specifically focusing on last names

oai_token-act-pair · gpt-3.5-turbo

names of individuals, particularly those involved in political or public discussions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 jumper

-0.77

 Butterfly

-0.73

 ribbon

-0.73

 Galaxy

-0.72

 kicker

-0.68

 conditioning

-0.66

 aspirin

-0.66

 heart

-0.65

 controlled

-0.65

 chill

-0.64

POSITIVE LOGITS

swers

1.07

romeda

1.06

yers

1.02

son

1.01

rick

0.98

assador

0.98

Bron

0.97

ards

0.96

rea

0.96

rical

0.95

Activations Density 0.067%

names related to people or entities, specifically focusing on last names

names of individuals, particularly those involved in political or public discussions

No Comments

No Known Activations