INDEX

Explanations

phrases indicating if an action or decision is justified

oai_token-act-pair · gpt-3.5-turbo

phrases related to the concept of justification or legitimacy

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

TL

-0.63

roid

-0.62

ore

-0.61

onna

-0.60

dll

-0.59

Kun

-0.59

 Processing

-0.58

 Destruction

-0.58

 Entry

-0.58

Lup

-0.57

POSITIVE LOGITS

 justified

3.65

 warranted

2.32

 justifies

2.23

 justifying

1.90

 justify

1.79

 justification

1.52

 deserved

1.28

 unjust

1.25

 reasoned

1.23

reasonable

1.18

Activations Density 0.012%

phrases indicating if an action or decision is justified

phrases related to the concept of justification or legitimacy

No Comments

No Known Activations