INDEX

Explanations

phrases and terms related to medical and clinical classifications

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

CR followed by number or marker

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

extAlignment

-0.65

toplankton

-0.63

dafx

-0.61

Portail

-0.60

 yPos

-0.58

ității

-0.57

 lenker

-0.57

getUserId

-0.56

Yang

-0.55

ETO

-0.55

POSITIVE LOGITS

CR

0.81

CR

0.80

Cr

0.76

Cr

0.73

cr

0.65

 Crim

0.59

cr

0.59

Cri

0.57

Cru

0.55

crv

0.55

Activations Density 0.053%

phrases and terms related to medical and clinical classifications

CR followed by number or marker

No Comments

No Known Activations