INDEX

Explanations

references to coding structures and components

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

programming code and testing

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

+#+

-0.55

原始内容存档于

-0.54

dafx

-0.53

iſten

-0.52

 Meksiku

-0.52

 vPvB

-0.51

oredCriteria

-0.51

 gainera

-0.51

niſſe

-0.50

Jereo

-0.50

POSITIVE LOGITS

RegressionTest

0.38

новништво

0.37

 Erben

0.36

 <<<<<<<<<<<<<<

0.34

ULLY

0.34

ModelForm

0.32

ILLON

0.32

 Hause

0.31

 δὲ

0.30

ully

0.30

Activations Density 0.004%

references to coding structures and components

programming code and testing

No Comments

No Known Activations