INDEX

Explanations

geographical references and directional terms

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

directions, exits, highways

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

DoubleQuotes

-0.41

 pleg

-0.39

 inhoud

-0.39

 cree

-0.38

 misure

-0.36

čaj

-0.35

 Análisis

-0.35

unanje

-0.35

 Norsk

-0.34

 crees

-0.33

POSITIVE LOGITS

 roads

0.56

 highways

0.55

 highway

0.52

drives

0.52

 freeway

0.49

highway

0.49

 drives

0.48

 traffic

0.48

 detour

0.48

pendicular

0.47

Activations Density 0.010%

geographical references and directional terms

directions, exits, highways

No Comments

No Known Activations