INDEX

Explanations

references to user interface design elements and layout functions

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

centered layout

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

CppCodeGen

-0.49

ThroughAttribute

-0.46

()))

-0.42

tagHelper

-0.42

mjs

-0.42

[])

-0.40

iD

-0.40

kle

-0.40

:✨

-0.39

 reconstru

-0.39

POSITIVE LOGITS

RelativeTo

2.72

 relative

0.73

relative

0.72

GraphicsUnit

0.68

 Relative

0.68

RELATIVE

0.63

IntoConstraints

0.63

Relative

0.60

 RELATIVE

0.60

relativeTime

0.55

Activations Density 0.000%

references to user interface design elements and layout functions

centered layout

No Comments

No Known Activations