INDEX

Explanations

phrases that refer to people or their identities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Literatuur

-0.44

 drep

-0.39

 détru

-0.38

 proprement

-0.38

:✨

-0.38

 materialen

-0.37

 sexuales

-0.37

 utveckling

-0.36

 testé

-0.35

 épu

-0.34

POSITIVE LOGITS

AxisAlignment

0.60

rungsseite

0.60

 nakalista

0.60

ISupport

0.58

تقاوى

0.57

RTHOOK

0.56

 समीक्षाओं

0.56

ArgumentParser

0.54

الدراسه

0.52

 Autorisations

0.52

Activations Density 0.017%

phrases that refer to people or their identities

No Comments

No Known Activations