INDEX

Explanations

references to "full" or complete entities, often in a compound form or leading a phrase

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

"full" and its variants

np_acts-logits-general · gemini-2.0-flash

full of something

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

EDEFAULT

-0.93

+#+

-0.81

 altogether

-0.71

 AssemblyProduct

-0.70

Efq

-0.69

WireFormatLite

-0.66

principalTable

-0.65

VersionUID

-0.64

 Wicidata

-0.64

 meille

-0.64

POSITIVE LOGITS

 full

1.16

 Full

1.16

Full

1.06

 FULL

1.02

FULL

1.00

fledged

1.00

full

0.97

 fled

0.90

 blown

0.88

blown

0.82

Activations Density 0.090%

references to "full" or complete entities, often in a compound form or leading a phrase

"full" and its variants

full of something

No Comments

No Known Activations