INDEX

Explanations

terms associated with health and medical conditions

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

specific compound terms

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

 hubieran

-0.43

 alrededores

-0.42

 quisiera

-0.41

 comigo

-0.41

icznie

-0.40

 lendemain

-0.40

antaranya

-0.38

 Tiefe

-0.37

 berdua

-0.37

somewhere

-0.37

POSITIVE LOGITS

RenderAtEndOf

0.66

 reef

0.59

AddTagHelper

0.59

 pecan

0.56

 apache

0.54

 milkshake

0.53

 arab

0.53

 milk

0.52

__":

0.52

 health

0.52

Activations Density 0.063%

terms associated with health and medical conditions

specific compound terms

No Comments

No Known Activations