INDEX

Explanations

superlatives or expressions denoting extreme qualities

oai_token-act-pair · gpt-3.5-turbo

the term "absolute" as it is used in various contexts, particularly to emphasize certainty or superiority

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

chan

-0.65

vernment

-0.62

sold

-0.61

arry

-0.60

wagen

-0.60

 helps

-0.60

 Doodle

-0.59

ries

-0.59

talk

-0.58

 Chem

-0.58

POSITIVE LOGITS

 absolute

3.56

absolute

2.80

 Absolute

2.14

 absolutely

1.43

 unconditional

1.43

relative

1.38

 utmost

1.35

 relative

1.30

 ultimate

1.22

 exact

1.20

Activations Density 0.011%

superlatives or expressions denoting extreme qualities

the term "absolute" as it is used in various contexts, particularly to emphasize certainty or superiority

No Comments

No Known Activations