INDEX

Explanations

references to secretive or encoded communication

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

ReusableCell

-0.45

 either

-0.39

 oprot

-0.37

 <<<<<<<<<<<<<<

-0.36

UserDao

-0.35

 来自

-0.35

:✨

-0.35

 quoted

-0.34

 cited

-0.34

UserController

-0.34

POSITIVE LOGITS

tvguidetime

0.57

 secret

0.56

secret

0.54

暗

0.52

 SECRET

0.52

 secreta

0.51

 secreto

0.50

Secret

0.50

 '\\;'

0.49

SECRET

0.48

Activations Density 0.035%

references to secretive or encoded communication

No Comments

No Known Activations