INDEX

Explanations

negative descriptors and criticisms related to content or performances

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.20.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

 TextAppearance

-0.40

 configure

-0.35

paced

-0.35

 noDo

-0.33

 الحره

-0.32

 diff

-0.32

 Missing

-0.31

configure

-0.31

 deferred

-0.31

 Forbidden

-0.31

POSITIVE LOGITS

 buruk

0.77

 poor

0.71

 horrible

0.69

 peores

0.67

 poorly

0.65

 worst

0.65

contentLoaded

0.65

 terrible

0.65

 horrendous

0.64

 awful

0.63

Activations Density 0.273%

negative descriptors and criticisms related to content or performances

No Comments

No Known Activations