INDEX

Explanations

phrases indicating inquiry or observation

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

words and phrases seeking information

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

-0.42

-0.40

ſelf

-0.39

 emperador

-0.37

The

-0.35

The

-0.35

 Houſe

-0.35

 inconvénients

-0.31

 Currently

-0.31

 dispositif

-0.31

POSITIVE LOGITS

 חיצוניים

0.81

rungsseite

0.78

 Geſch

0.73

 oprot

0.71

mbgg

0.71

 müſſen

0.69

ſſung

0.66

iſchen

0.66

 שוליים

0.65

 verſch

0.65

Activations Density 0.073%

phrases indicating inquiry or observation

words and phrases seeking information

No Comments

No Known Activations