INDEX

Explanations

terms related to previous professional roles and titles

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

SCN

-0.45

 Endlich

-0.38

 grows

-0.37

ASG

-0.37

!***

-0.37

HostException

-0.36

Assured

-0.36

HBO

-0.36

ocations

-0.35

hci

-0.35

POSITIVE LOGITS

+#+

0.50

 intptr

0.48

 мәкал

0.47

المناصب

0.47

 мәкалә

0.46

IBOutlet

0.45

 Administrativna

0.44

synthesize

0.44

 Signalez

0.44

abstractmethod

0.44

Activations Density 0.008%

terms related to previous professional roles and titles

No Comments

No Known Activations