INDEX

Explanations

coding patterns and structures

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 19-gemmascope-mlp-16k

Configuration

google/gemma-scope-2b-pt-mlp/layer_19/width_16k/average_l0_109

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.19.hook_mlp_out

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 الرياضيه

-0.62

 InputDecoration

-0.62

ViewFeatures

-0.59

Véxase

-0.58

doria

-0.57

ParallelGroup

-0.56

 سكانية

-0.56

APOLIS

-0.56

 насељу

-0.54

LOTTE

-0.54

POSITIVE LOGITS

 noqa

0.54

icylic

0.50

 littéraire

0.50

توض

0.48

)))));

0.46

 sabbia

0.46

 Starting

0.46

+:+

0.45

 starting

0.44

 difficoltà

0.44

Activations Density 0.163%

coding patterns and structures

No Known Activations