INDEX

Explanations

phrases indicating justification or rationale

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

reason to do

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

providedIn

-0.35

 vaisselle

-0.33

 objectMapper

-0.30

 linguistique

-0.28

 Serikat

-0.28

 réservé

-0.27

jamento

-0.27

 processes

-0.27

 Américas

-0.27

 Atlántico

-0.27

POSITIVE LOGITS

 reason

0.84

why

0.83

rungsseite

0.81

Reasons

0.80

 Reasons

0.79

reason

0.79

 REASONS

0.77

why

0.76

 Reason

0.73

reasons

0.72

Activations Density 0.034%

phrases indicating justification or rationale

reason to do

No Comments

No Known Activations