INDEX

Explanations

the presence of the word "not" in various contexts

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

'>{

-0.61

})->

-0.54

 ویکی‌پدی

-0.53

>'+

-0.51

ductory

-0.51

€/

-0.49

terase

-0.49

ViewChild

-0.49

lccn

-0.47

'),

-0.47

POSITIVE LOGITS

Anyways

0.60

omitempty

0.54

SqlClient

0.49

GraphicsUnit

0.49

imanapun

0.48

 Anyways

0.47

Cordialement

0.45

Anyhow

0.45

Whether

0.45

Activations Density 0.006%

the presence of the word "not" in various contexts

No Comments

No Known Activations