INDEX

Explanations

terms related to secrecy and confidential information

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

New Auto-Interp

Top Features by Cosine Similarity

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

Configuration

google/gemma-scope-9b-pt-res/layer_2/width_16k/average_l0_67

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.2.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

 للاسماء

-0.50

bootstrapcdn

-0.50

assertRaises

-0.48

 فريبيس

-0.47

#+#

-0.47

kkelijk

-0.45

:+:

-0.44

AddTagHelper

-0.44

pensive

-0.43

itinéraire

-0.43

POSITIVE LOGITS

 secret

0.85

 Secret

0.79

Secret

0.76

 story

0.70

secret

0.70

 secreto

0.64

 SECRET

0.64

 secrets

0.61

 segredo

0.60

 extreme

0.59

Activations Density 0.058%

terms related to secrecy and confidential information

No Comments

No Known Activations