INDEX

Explanations

verbs related to action and enforcement

oai_token-act-pair · gpt-3.5-turbo

action verbs related to management and intervention

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 Extrem

-0.67

>>\

-0.65

 msec

-0.64

utenberg

-0.62

 Ruins

-0.56

perty

-0.56

mur

-0.54

 Floor

-0.54

 fracturing

-0.53

 Anger

-0.53

POSITIVE LOGITS

able

1.09

ments

1.07

ables

1.01

ings

1.00

ability

1.00

 accordingly

0.99

ingly

0.97

ably

0.94

 versa

0.92

ible

0.85

Activations Density 0.385%

verbs related to action and enforcement

action verbs related to management and intervention

No Known Activations