INDEX

Explanations

references to negative qualities or situations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 2-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_2/width_16k/average_l0_67

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.2.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 heavily

-0.47

 highly

-0.45

Highly

-0.43

 strongly

-0.43

highly

-0.42

 Highly

-0.41

 Dead

-0.40

ÊN

-0.39

 handover

-0.38

 fiercely

-0.38

POSITIVE LOGITS

bad

3.13

bad

2.69

Bad

2.61

Bad

2.52

BAD

2.34

BAD

2.30

坏

1.46

 schlechte

1.41

 malos

1.31

 mauvaise

1.30

Activations Density 0.060%

references to negative qualities or situations

No Comments

No Known Activations