INDEX

Explanations

informal language and expressions of personal opinions

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

haven, isn, don, wouldn

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

 šķ

-0.54

 Gegenteil

-0.46

 precedentes

-0.42

 pungkas

-0.41

 siitä

-0.40

 aislados

-0.40

 apsau

-0.39

 iniciativa

-0.39

 Italij

-0.38

 själva

-0.38

POSITIVE LOGITS

bibfield

0.60

 ſehen

0.59

ſicht

0.59

BibitemShut

0.57

 zwiſchen

0.57

 Menſchen

0.57

<unused14>

0.56

<unused28>

0.56

<unused3>

0.56

[@BOS@]

0.56

Activations Density 0.117%

informal language and expressions of personal opinions

haven, isn, don, wouldn

No Comments

No Known Activations