INDEX

Explanations

references to specific numerical values or counts in a variety of contexts

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

1

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

Final

-0.39

WEB

-0.35

Tru

-0.35

Во

-0.35

-------------</

-0.35

Adapun

-0.34

 nahilalakip

-0.34

 propi

-0.34

Pass

-0.34

原标题

-0.34

POSITIVE LOGITS

ViewFeatures

0.58

Alike

0.55

 TAMBIÉN

0.54

 onOptions

0.53

 дописавши

0.52

MessageTagHelper

0.50

 ThemeData

0.50

expandindo

0.50

 kasarigan

0.49

ThroughAttribute

0.49

Activations Density 0.090%

references to specific numerical values or counts in a variety of contexts

1

No Comments

No Known Activations