INDEX

Explanations

conjunctive phrases that connect ideas or clauses, particularly variations of "and."

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

and followed by diverse words

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

 Männern

-0.46

 nemlig

-0.43

guien

-0.42

 azonban

-0.42

 تضيفلها

-0.41

 čia

-0.40

 totiž

-0.39

 Corée

-0.39

 Suomessa

-0.39

 Paglinawan

-0.38

POSITIVE LOGITS

الحياه

0.59

 ujednoznacz

0.58

orek

0.54

észetes

0.52

stus

0.52

idig

0.51

rog

0.51

PRS

0.50

ibase

0.49

fod

0.49

Activations Density 0.077%

conjunctive phrases that connect ideas or clauses, particularly variations of "and."

and followed by diverse words

No Comments

No Known Activations