INDEX

Explanations

specific codes and references commonly used in documentation and data structures

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

Numbers in ranges, especially "1 through" or "to"

np_acts-logits-general · gemini-2.0-flash

sequences like 1 through 5

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

 AssemblyCulture

-0.91

Personendaten

-0.85

']],

-0.76

()].

-0.75

WebVitals

-0.74

()]

-0.71

tableFuture

-0.71

 Normdatei

-0.71

帖最后由

-0.69

')],

-0.69

POSITIVE LOGITS

 through

2.17

through

1.87

Through

1.83

 Through

1.76

 THROUGH

1.75

 thru

1.74

THROUGH

1.46

 attraverso

1.34

thru

1.31

 Thru

1.25

Activations Density 1.192%

specific codes and references commonly used in documentation and data structures

Numbers in ranges, especially "1 through" or "to"

sequences like 1 through 5

No Comments

No Known Activations