INDEX

Explanations

patterns related to code elements and structures, such as function calls and conditional statements

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

Characters often used in programming/code

np_acts-logits-general · gemini-2.0-flash

code delimiters and programming keywords

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 24-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_24/width_16k/average_l0_73

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.24.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

key

-0.77

pre

-0.77

set

-0.77

new

-0.77

 fore

-0.76

 number

-0.75

add

-0.73

 this

-0.73

 model

-0.72

 post

-0.72

POSITIVE LOGITS

 volontà

0.61

 tradição

0.57

 consciência

0.56

 mauvaises

0.55

 élevées

0.54

 violência

0.54

 casada

0.54

 voluntad

0.54

 testigos

0.53

 către

0.53

Activations Density 0.768%

patterns related to code elements and structures, such as function calls and conditional statements

Characters often used in programming/code

code delimiters and programming keywords

No Comments

No Known Activations