INDEX

Explanations

verbs or actions related to processing and modification

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

removal

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

 SwitchCompat

-0.45

✭✭

-0.45

RunWith

-0.39

}{@

-0.38

 argint

-0.38

ติด

-0.36

onAttach

-0.35

AndroidJUnit

-0.35

 المعيارى

-0.35

ærk

-0.34

POSITIVE LOGITS

 removal

0.72

removal

0.60

 clean

0.60

cleaner

0.59

 removing

0.59

 remove

0.58

 Removal

0.58

 removals

0.57

 cleaning

0.57

 cleaner

0.56

Activations Density 0.104%

verbs or actions related to processing and modification

removal

No Comments

No Known Activations