INDEX

Explanations

calculations and values related to experimental results and data analysis

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

calculation by formula

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

 للاسماء

-0.60

الحياه

-0.56

 nakalista

-0.47

anke

-0.47

 оригіналу

-0.46

〢

-0.45

Portály

-0.45

CppCodeGen

-0.44

FTFY

-0.43

laise

-0.43

POSITIVE LOGITS

 przep

0.45

 ratio

0.43

 measures

0.40

 grade

0.38

 användas

0.38

 measure

0.37

 bestaan

0.36

 calculate

0.36

 czł

0.35

ConverterFactory

0.35

Activations Density 0.147%

calculations and values related to experimental results and data analysis

calculation by formula

No Comments

No Known Activations