INDEX

Explanations

tokens that signal the beginning of a structured format or data entry, such as lists or selections

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

Code snippets, locations, or XML-like syntax

np_acts-logits-general · gemini-2.0-flash

archaic spellings and punctuation

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 22-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_22/width_16k/average_l0_72

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.22.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

-1.15

-1.09

-0.89

<eos>

-0.89

↵↵

-0.82

-0.80

-0.78

-0.76

-0.74

POSITIVE LOGITS

 itſelf

1.39

 pleaſure

1.36

 myſelf

1.30

 purpoſe

1.25

 raiſ

1.23

 ſever

1.20

$_"

1.19

 faſt

1.19

 reaſon

1.18

 ―――――

1.17

Activations Density 0.193%

tokens that signal the beginning of a structured format or data entry, such as lists or selections

Code snippets, locations, or XML-like syntax

archaic spellings and punctuation

No Comments

No Known Activations