INDEX

Explanations

references to memory address operations using the ampersand symbol

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

uxxxx

-0.54

 enfans

-0.52

 canst

-0.52

isContained

-0.52

ロウィン

-0.52

othelioma

-0.51

 scolaires

-0.50

<unused0>

-0.50

olverine

-0.50

chieht

-0.50

POSITIVE LOGITS

(&

0.91

 &___

0.69

(&(

0.60

(&$

0.60

WriteTagHelper

0.57

>(&

0.56

((&

0.55

(&_

0.54

+#+#

0.54

*)&

0.53

Activations Density 0.043%

references to memory address operations using the ampersand symbol

No Comments

No Known Activations