INDEX

Explanations

quantitative estimates and measurements with approximate values

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

approximation of quantities or times

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

 nemlig

-0.39

 absolutamente

-0.36

 absolutely

-0.36

 jednocześnie

-0.35

 türlü

-0.34

insatz

-0.34

 obě

-0.33

ajemen

-0.32

 oorlog

-0.32

 refuser

-0.31

POSITIVE LOGITS

GEBURTSDATUM

0.88

 approximate

0.81

 Approximate

0.80

approx

0.78

 approx

0.76

ConstraintMaker

0.75

Approximate

0.75

 Approx

0.75

approximate

0.71

Approx

0.71

Activations Density 0.196%

quantitative estimates and measurements with approximate values

approximation of quantities or times

No Comments

No Known Activations