INDEX

Explanations

phrases related to programming logic and execution flow

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 nacionais

-0.49

ThroughAttribute

-0.47

 Umgang

-0.46

 redenen

-0.44

protoimpl

-0.43

 toimi

-0.43

 Мексичка

-0.42

 Roskov

-0.42

 betreft

-0.42

กัน

-0.42

POSITIVE LOGITS

 gravity

0.71

 Somehow

0.68

Somehow

0.67

 Résultat

0.66

 AssemblyCulture

0.65

 udaler

0.62

DECREF

0.62

 transfieras

0.62

 when

0.61

وقتی

0.61

Activations Density 0.651%

phrases related to programming logic and execution flow

No Comments

No Known Activations