INDEX

Explanations

patterns in numerical data or sequences

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 13-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_13/width_65k/average_l0_75

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.13.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

AnimationsModule

-0.47

<eos>

-0.44

 Дата

-0.44

able

-0.42

一身

-0.40

Top

-0.40

 qualitative

-0.39

上一

-0.39

Durée

-0.38

exo

-0.38

POSITIVE LOGITS

RegressionTest

0.82

 الرياضيه

0.77

 الحره

0.71

 propOrder

0.68

الدراسه

0.67

InitVars

0.67

 doInBackground

0.66

談社

0.66

 betweenstory

0.66

tvguidetime

0.65

Activations Density 0.003%

patterns in numerical data or sequences

No Known Activations