INDEX

Explanations

occurrences of specific zero-activation elements within a structured context

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

technical or code identifiers

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

 Masyarakat

-0.47

 Jurí

-0.46

+:+

-0.46

 jurídico

-0.44

nalpot

-0.43

 dourada

-0.43

 Übungen

-0.42

wym

-0.42

endpush

-0.42

rozco

-0.41

POSITIVE LOGITS

rungsseite

0.59

bootstrapcdn

0.55

 AssemblyCulture

0.51

ItemBackground

0.40

MetaObject

0.40

 gynhyrchwyd

0.40

ModelAdmin

0.39

StateToProps

0.38

 prop

0.38

'{@

0.38

Activations Density 0.453%

occurrences of specific zero-activation elements within a structured context

technical or code identifiers

No Comments

No Known Activations