INDEX

Explanations

references to conferences and their respective details

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

ordinal followed by conference or award

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

 podamos

-0.34

dAtA

-0.33

[]:

-0.33

 secours

-0.33

 sólido

-0.32

 sorpresas

-0.31

 sobě

-0.31

ไว้

-0.31

 mínima

-0.31

 priorité

-0.30

POSITIVE LOGITS

 незавершена

0.65

writeField

0.63

rungsseite

0.61

AndEndTag

0.59

parab

0.58

 Савезне

0.56

:✨

0.56

Aholisi

0.53

symmetric

0.52

underwater

0.52

Activations Density 0.029%

references to conferences and their respective details

ordinal followed by conference or award

No Comments

No Known Activations