INDEX

Explanations

file directory paths and executable commands

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

executables in /bin

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

iſen

-0.66

المكان

-0.65

majánló

-0.62

 Verſ

-0.59

transQ

-0.59

 zoude

-0.59

enablog

-0.59

 ſind

-0.59

tagHelperRunner

-0.59

niſſe

-0.59

POSITIVE LOGITS

0.46

sbin

0.43

bin

0.42

0.40

 electric

0.39

interopRequire

0.39

Mik

0.38

="

0.37

bin

0.36

 Bill

0.36

Activations Density 0.004%

file directory paths and executable commands

executables in /bin

No Comments

No Known Activations