INDEX

Explanations

specific patterns and structures in numerical or tabular data

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

Occurrences before "Table" in filenames

np_acts-logits-general · gemini-2.0-flash

table followed by number

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

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

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 nahilalakip

-0.70

 للمعارف

-0.63

 houſe

-0.62

AndEndTag

-0.57

NameInMap

-0.54

########.

-0.53

 fuper

-0.51

 pleaſure

-0.51

 cauſe

-0.50

ſelf

-0.49

POSITIVE LOGITS

 eksemp

0.40

 propOrder

0.39

tå

0.38

 IFTT

0.36

IRQn

0.36

 manquer

0.34

 Schmid

0.33

 חיצוניים

0.32

GraphicsUnit

0.32

 nôtre

0.32

Activations Density 0.478%

specific patterns and structures in numerical or tabular data

Occurrences before "Table" in filenames

table followed by number

No Comments

No Known Activations