INDEX

Explanations

phrases related to power or strength

oai_token-act-pair · gpt-3.5-turbo

expressions related to power dynamics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

sho

-0.69

je

-0.67

 decorated

-0.67

 peanuts

-0.62

 pneumonia

-0.61

che

-0.61

 escorted

-0.60

 cakes

-0.59

ADA

-0.59

 distant

-0.58

POSITIVE LOGITS

power

4.20

Power

2.10

powers

2.09

powered

1.95

 power

1.66

 POWER

1.66

 Power

1.65

energy

1.26

force

1.25

volt

1.18

Activations Density 0.006%

phrases related to power or strength

expressions related to power dynamics

No Comments

No Known Activations