INDEX

Explanations

instances of doors and hatches being opened or closed

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

open position

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

 kasarigan

-0.53

stylers

-0.49

 nimic

-0.48

 للمعارف

-0.47

 plă

-0.47

 Standort

-0.47

 переписи

-0.47

 zoude

-0.47

打ち

-0.46

 argint

-0.46

POSITIVE LOGITS

 open

0.98

 opened

0.96

open

0.87

 Opened

0.86

 opening

0.86

 OPEN

0.85

 Opening

0.85

 opens

0.82

opened

0.82

opening

0.81

Activations Density 0.028%

instances of doors and hatches being opened or closed

open position

No Comments

No Known Activations