INDEX

Explanations

terms related to excessive behavior or self-gratification

oai_token-act-pair · gpt-3.5-turbo

terms related to indulgence and arbitration

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

è¦ļéĨĴ

-0.72

 Doctrine

-0.70

 Biological

-0.69

 Shack

-0.68

 Murd

-0.67

Doe

-0.66

IELD

-0.64

 Butterfly

-0.64

beard

-0.64

 Waterloo

-0.62

POSITIVE LOGITS

rative

1.17

rator

1.08

ging

1.07

ration

1.05

acy

0.96

ates

0.96

gent

0.95

iments

0.92

rators

0.91

gregation

0.91

Activations Density 0.043%

terms related to excessive behavior or self-gratification

terms related to indulgence and arbitration

No Comments

No Known Activations