INDEX

Explanations

expressions of potential outcomes and actions

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 leaſt

-0.44

 skjer

-0.40

 lösen

-0.36

 personales

-0.36

 süß

-0.36

 interactiva

-0.35

 ſtate

-0.35

 privados

-0.35

 szczegół

-0.35

 hiasan

-0.34

POSITIVE LOGITS

 utafitiHapana

0.67

 Biôgrafia

0.55

AddHtmlAttribute

0.54

 Autorizaciones

0.54

dafx

0.52

 jadx

0.52

kháu

0.51

✨:

0.50

windowFixed

0.50

TargetApi

0.50

Activations Density 0.048%

expressions of potential outcomes and actions

No Comments

No Known Activations