INDEX

Explanations

references to identical or similar names and definitions across various contexts

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

same name or identifier

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

Mor

-0.43

 Traffic

-0.40

 Holz

-0.35

 Wikip

-0.35

 shorten

-0.34

mor

-0.34

 Jakob

-0.34

ˈ

-0.33

 program

-0.33

Traffic

-0.33

POSITIVE LOGITS

homonymie

0.64

 disambiguazione

0.60

ſelf

0.54

 виправивши

0.52

AddTagHelper

0.52

 otomatig

0.51

 cauſe

0.50

wiſe

0.49

localctx

0.49

EndContext

0.48

Activations Density 0.217%

references to identical or similar names and definitions across various contexts

same name or identifier

No Comments

No Known Activations