INDEX

Explanations

comparative terms that denote improvement or superiority

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 38-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_38/width_16k/average_l0_128

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.38.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

 enough

-0.66

 Enough

-0.60

enough

-0.59

Enough

-0.53

 ENOUGH

-0.53

fulness

-0.52

 genug

-0.46

sufficient

-0.45

 weitere

-0.44

️

-0.43

POSITIVE LOGITS

 than

1.27

 THAN

1.04

THAN

1.02

Than

1.00

 niż

0.95

than

0.91

 Than

0.79

 než

0.66

 decât

0.62

 kuin

0.59

Activations Density 0.177%

comparative terms that denote improvement or superiority

No Comments

No Known Activations