INDEX

Explanations

phrases related to personal preferences and emotions

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

mathematical or programming terms

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

 comienzan

-0.35

 بگیر

-0.34

Sklici

-0.31

 HttpHeaders

-0.31

 participan

-0.31

 comenzado

-0.30

场比赛

-0.30

 empiezan

-0.30

 instead

-0.30

 autobuses

-0.30

POSITIVE LOGITS

Geplaatst

0.70

 <>",

0.59

 AssemblyCompany

0.55

casio

0.55

AutoScaleMode

0.54

writeFieldEnd

0.53

casila

0.53

AndroidJUnit

0.52

internalType

0.51



0.49

Activations Density 0.067%

phrases related to personal preferences and emotions

mathematical or programming terms

No Comments

No Known Activations