INDEX

Explanations

comparative and evaluative phrases

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

as expected or desired

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

ahren

-0.56

AddTagHelper

-0.49

odis

-0.47

comfortably

-0.45

oria

-0.44

laring

-0.44

EMBER

-0.43

tms

-0.43

 الترك

-0.43

lares

-0.42

POSITIVE LOGITS

 expected

0.49

 conmigo

0.43

 deseado

0.42

SharedCtor

0.41

 informée

0.41

 desired

0.40

 sonhos

0.39

 comigo

0.39

预期

0.37

 esperado

0.37

Activations Density 0.057%

comparative and evaluative phrases

as expected or desired

No Comments

No Known Activations