INDEX

Explanations

alphanumeric characters and patterns, possibly related to coding or data structures

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

start of user input

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

 surla

-0.71

tagHelperRunner

-0.64

expandindo

-0.60

RegressionTest

-0.59

<unused28>

-0.58

 utafitiHapana

-0.58

<unused14>

-0.57

<unused43>

-0.57

<unused8>

-0.57

<unused7>

-0.57

POSITIVE LOGITS

 snart

0.33

 échanges

0.33

’

0.30

 dupla

0.30

Soon

0.29

 reunião

0.29

 puits

0.29

WriteBarrier

0.28

 regione

0.28

 soon

0.28

Activations Density 0.002%

alphanumeric characters and patterns, possibly related to coding or data structures

start of user input

No Comments

No Known Activations