INDEX

Explanations

words related to causing a specific effect or state in a strong and impactful manner

oai_token-act-pair · gpt-3.5-turbo

terms related to the act of inducing or causing specific effects or conditions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Ü

-0.69

 Chop

-0.68

foundation

-0.67

fman

-0.65

 Shattered

-0.65

aldo

-0.65

 Responsibility

-0.64

esides

-0.64

apest

-0.64

iffin

-0.63

POSITIVE LOGITS

 induce

1.01

 induced

1.00

 inducing

0.96

induced

0.92

 induces

0.91

uces

0.91

uced

0.88

ocy

0.88

 induction

0.82

 cele

0.82

Activations Density 0.013%

words related to causing a specific effect or state in a strong and impactful manner

terms related to the act of inducing or causing specific effects or conditions

No Comments

No Known Activations