INDEX

Explanations

negations and commands

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 7-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_7/width_16k/average_l0_92

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.7.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

 Chwiliwch

-0.75

transQ

-0.72

SequentialGroup

-0.69

tagHelperRunner

-0.69

 ब्रेकडाउन

-0.68

 beginnetje

-0.66

RegressionTest

-0.64

enterOuterAlt

-0.63

tvguidetime

-0.63

 Мексичка

-0.63

POSITIVE LOGITS

 Jangan

0.66

 đừng

0.60

Jangan

0.60

DoNot

0.58

 jangan

0.57

Dont

0.56

 不要

0.56

 Dont

0.53

dont

0.50

你不要

0.47

Activations Density 0.109%

negations and commands

No Comments

No Known Activations