INDEX

Explanations

bracketed expressions and shape-related terms in programming context

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

code structure and programming constructs

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

 Administrativna

-0.50

kWh

-0.42

IndentedString

-0.41

पया

-0.41

kong

-0.41

 kong

-0.40

 Ponto

-0.39

naje

-0.38

chaun

-0.38

tagext

-0.38

POSITIVE LOGITS

 GenerationType

0.51

zzleHttp

0.50

ginate

0.47

SaveChangesAsync

0.46

 Zusammensetzung

0.43

BagLayout

0.43

ScopeManager

0.43

 conjunta

0.41

 wöch

0.41

lī

0.40

Activations Density 0.002%

bracketed expressions and shape-related terms in programming context

code structure and programming constructs

No Comments

No Known Activations