INDEX

Explanations

references to data processing and circuit design

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

conventional circuit problems

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

WriteBarrier

-0.64

contentLoaded

-0.63

 Biôgrafia

-0.60

исленность

-0.58

帖最后由

-0.58

 "..\..\..\

-0.57

 "..\..\

-0.56

 Мексичка

-0.56

ksikon

-0.54

NUMX

-0.52

POSITIVE LOGITS

 оригіналу

0.41

太过

0.39

原来的

0.33

 Freundin

0.32

太

0.31

 vieill

0.30

ji

0.30

 invokingState

0.30

useState

0.30

 terlalu

0.29

Activations Density 0.404%

references to data processing and circuit design

conventional circuit problems

No Comments

No Known Activations