INDEX

Explanations

phrases related to losing or winning in a competitive context

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

location or event descriptions

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

 navíc

-0.38

としては

-0.38

 integridad

-0.37

ílica

-0.35

 navigable

-0.35

 içerisinde

-0.35

хоз

-0.35

 sartén

-0.33

もありました

-0.33

 vše

-0.33

POSITIVE LOGITS

 تضيفلها

0.61

 Numerade

0.53

WriteTagHelper

0.52

 getContext

0.51

 FetchType

0.51

OGND

0.50

ValueStyle

0.49

ьаж

0.49

CppMethod

0.48

 ſch

0.47

Activations Density 0.040%

phrases related to losing or winning in a competitive context

location or event descriptions

No Comments

No Known Activations