INDEX

Explanations

sequences of whitespace characters in text

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

code structures and programming keywords

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-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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

uxxxx

-0.75

 OMITBAD

-0.53

 ErrIntOverflow

-0.52

 <<<<<<<<<<<<<<

-0.51

angelog

-0.49

wpi

-0.48

 vervangen

-0.48

 HFILL

-0.46

harapkan

-0.46

fillColor

-0.46

POSITIVE LOGITS

DockStyle

0.45

borderSide

0.43

 GenerationType

0.42

 irra

0.41

 corona

0.41

AlterField

0.39

tomans

0.39

 arbit

0.38

 ***!

0.37

MigrationBuilder

0.37

Activations Density 0.005%

sequences of whitespace characters in text

code structures and programming keywords

No Comments

No Known Activations