INDEX

Explanations

modal verbs and their negative forms indicating possibility or likelihood

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

 AssemblyCompany

-0.65

 Gunakan

-0.49

setVerticalGroup

-0.47

 Tinggi

-0.46

ToUse

-0.45

ondere

-0.44

istungs

-0.44

 phóng

-0.43

 Región

-0.43

forget

-0.42

POSITIVE LOGITS

 helps

0.63

 brings

0.59

 enables

0.57

 ensures

0.55

 help

0.54

 bring

0.53

 lead

0.51

 leads

0.51

 causes

0.50

complexContent

0.50

Activations Density 0.517%

modal verbs and their negative forms indicating possibility or likelihood

No Comments

No Known Activations