INDEX

Explanations

commands or instructions directing specific actions or behaviors

oai_token-act-pair · gpt-3.5-turbo

commands or actions directed at others

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

 strikingly

-0.72

requent

-0.71

 relatively

-0.69

 largely

-0.69

 purported

-0.68

 Particip

-0.67

ustomed

-0.67

 markedly

-0.66

 supplemented

-0.66

umerous

-0.65

POSITIVE LOGITS

me

1.47

ya

1.30

him

1.18

 yourselves

1.12

us

1.12

my

1.06

 thee

1.04

 daddy

1.03

HIM

1.03

 fuckin

1.03

Activations Density 0.523%

commands or instructions directing specific actions or behaviors

commands or actions directed at others

No Comments

No Known Activations