INDEX

Explanations

capitalized and abbreviated words or phrases, like legal or bureaucratic jargon, often containing numbers

oai_token-act-pair · gemini-2.0-flash

Academic/legal documents

np_max-act-logits · gemini-2.0-flash

New Auto-Interp

Configuration

google/gemma-scope-2b-pt-transcoders/layer_25/width_16k/average_l0_41

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.25.ln2.hook_normalized

Architecture

jumprelu_transcoder

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 itſelf

-1.08

 ſche

-1.08

ſelf

-1.04

 myſelf

-1.02

 abbildung

-1.02

 auffi

-1.02

 themſelves

-1.02

 pleaſure

-1.01

ſelves

-1.00

 purpoſe

-0.98

POSITIVE LOGITS

lu

0.52

hu

0.50

Ze

0.50

sat

0.49

sa

0.49

bu

0.49

sin

0.49

ho

0.48

cas

0.48

lit

0.48

Activations Density 6.255%

capitalized and abbreviated words or phrases, like legal or bureaucratic jargon, often containing numbers

Academic/legal documents

No Comments

No Known Activations