INDEX

Explanations

programmatic constructs related to object types and generics

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

object and pointer types

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

tvguidetime

-0.48

CloseOperation

-0.41

DockStyle

-0.40

 Drought

-0.38

 chron

-0.35

 Heading

-0.35

pagation

-0.35

плуата

-0.34

vantages

-0.34

 Deputy

-0.34

POSITIVE LOGITS

 jsPsych

0.53

whatever

0.51

 apapun

0.49

 виправивши

0.49

 quelcon

0.49

 feroit

0.47

agnos

0.46

汎

0.46

 whatever

0.46

generic

0.45

Activations Density 0.032%

programmatic constructs related to object types and generics

object and pointer types

No Comments

No Known Activations