INDEX

Explanations

phrases that express agreement or affirmation, particularly those starting with "Well."

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 1-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_1/width_16k/average_l0_69

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.1.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 disc

-0.41

ges

-0.40

bigsqcup

-0.40

 proper

-0.39

Nic

-0.38

ode

-0.37

Disc

-0.37

 route

-0.36

 Profes

-0.36

 métier

-0.36

POSITIVE LOGITS

parsedMessage

0.68

Speak

0.66

 speak

0.65

 Speak

0.62

SharedDtor

0.60

tell

0.60

speak

0.59

 betweenstory

0.58

 voeten

0.58

 feed

0.57

Activations Density 0.255%

phrases that express agreement or affirmation, particularly those starting with "Well."

No Known Activations