INDEX

Explanations

names of people, organizations, and software products

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

names and app titles

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

 незавершена

-0.76

thmetic

-0.60

 OMITBAD

-0.55

MemoryWarning

-0.54

TRAILING

-0.54

SharedDtor

-0.50

erdere

-0.50

brides

-0.49

 noastre

-0.49

 împre

-0.49

POSITIVE LOGITS

endregion

0.35

zydent

0.35

 vela

0.34

 interv

0.34

 colo

0.32

 lesen

0.31

dafri

0.31

hel

0.31

 centen

0.30

InstrumentedTest

0.30

Activations Density 0.080%

names of people, organizations, and software products

names and app titles

No Comments

No Known Activations