INDEX

Explanations

instances of negative or problematic contexts within text

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

data-label attributes

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

 enfans

-0.63

 étoit

-0.60

 pleaſure

-0.57

 pouvoit

-0.56

 feroit

-0.54

jenigen

-0.51

 JpaRepository

-0.50

 auroit

-0.49

 increí

-0.48

 dalamnya

-0.48

POSITIVE LOGITS

0.75

‐

0.59

://

0.58

<bos>

0.58

 Dana

0.57

__*/

0.57

 Proven

0.57

$_['

0.56

 Dyna

0.56

$-

0.56

Activations Density 0.004%

instances of negative or problematic contexts within text

data-label attributes

No Comments

No Known Activations