INDEX

Explanations

phrases that inquire about methods or approaches

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 without

-0.43

 recurrir

-0.42

addPreferredGap

-0.39

 Schmitz

-0.39

 Klage

-0.38

 vermek

-0.36

 ohne

-0.36

erequisites

-0.36

 Dekker

-0.36

 היתה

-0.35

POSITIVE LOGITS

DockStyle

0.65

المناصب

0.59

 people

0.56

OGND

0.55

 itſelf

0.54

twimg

0.52

 Monfieur

0.52

rungsseite

0.52

 humans

0.52

 Jefus

0.52

Activations Density 0.028%

phrases that inquire about methods or approaches

No Comments

No Known Activations