INDEX

Explanations

words related to ethical or moral judgments

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

before suffixes (-ys, -bent, -ized, -ance, -iness)

np_acts-logits-general · gemini-2.0-flash

foreign and proper names

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 22-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_22/width_16k/average_l0_72

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.22.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

ه

-1.38

te

-1.12

ta

-1.01

ia

-1.00

an

-0.97

на

-0.96

ي

-0.93

os

-0.89

as

-0.88

es

-0.86

POSITIVE LOGITS

neté

0.63

 disponibilités

0.63

hoeddwyd

0.60

 nationaux

0.58

 kasarigan

0.53

 Roskov

0.51

quakes

0.50

 prosegu

0.49

 Moab

0.49

GEBURTS

0.48

Activations Density 0.504%

words related to ethical or moral judgments

before suffixes (-ys, -bent, -ized, -ance, -iness)

foreign and proper names

No Comments

No Known Activations