INDEX

Explanations

various types of numerical or symbolic patterns, likely related to data formatting or extraction

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

filenames and paths with delimiters

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

 MenuItem

-0.32

WriteTagHelper

-0.31

每次

-0.31

 colors

-0.30

farben

-0.30

 testCase

-0.30

mejores

-0.29

 jadx

-0.29

iya

-0.29

 kleuren

-0.29

POSITIVE LOGITS

complexContent

0.75

ConstraintMaker

0.68

GEBURTS

0.67

 propOrder

0.61

asteroide

0.60

تقاوى

0.59

 betweenstory

0.59

 consultato

0.58

ArgsConstructor

0.58

 gynhyrchwyd

0.56

Activations Density 0.001%

various types of numerical or symbolic patterns, likely related to data formatting or extraction

filenames and paths with delimiters

No Comments

No Known Activations