INDEX

Explanations

variations of the word "people."

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

Token "pe" and variations

np_acts-logits-general · gemini-2.0-flash

pepsi, peoples, peeping

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

 Normdatei

-0.84

extAlignment

-0.80

__*/

-0.71

 InputDecoration

-0.71

 whiteColor

-0.66

NOPQRST

-0.65

(@"%@",

-0.65

uação

-0.62

Complete

-0.62

 Complete

-0.62

POSITIVE LOGITS

Pe

0.62

Pe

0.61

pe

0.56

 lâng

0.56

opel

0.47

 seaborn

0.47

haps

0.47

 ople

0.46

 peels

0.41

 пе

0.41

Activations Density 0.101%

variations of the word "people."

Token "pe" and variations

pepsi, peoples, peeping

No Comments

No Known Activations