INDEX

Explanations

details about architectural features and ornamentation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 21-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_21/width_16k/average_l0_129

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.21.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

 eût

-0.43

 étoient

-0.41

 Réponses

-0.40

paw

-0.40

 vícti

-0.39

 پای

-0.38

Према

-0.38

anken

-0.38

RTRS

-0.37

 hugging

-0.36

POSITIVE LOGITS

 roof

1.07

 Roof

0.96

 roofs

0.91

Roof

0.89

 roofing

0.88

 ROOF

0.84

roof

0.79

 header

0.74

 Roofing

0.72

 ceiling

0.71

Activations Density 0.678%

details about architectural features and ornamentation

No Comments

No Known Activations