INDEX

Explanations

technical terms and actions related to mathematical methods and problem-solving

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 17-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_17/width_65k/average_l0_125

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.17.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

 EconPapers

-0.50

 centaines

-0.46

posedge

-0.46

haupt

-0.45

dragen

-0.44

 daran

-0.43

gran

-0.42

osto

-0.42

fortawesome

-0.42

mappedBy

-0.40

POSITIVE LOGITS

 also

1.13

 therefore

1.12

 moreover

1.04

 nevertheless

0.99

 meanwhile

0.98

 indeed

0.91

 somehow

0.89

 nonetheless

0.88

 thus

0.87

 again

0.86

Activations Density 1.031%

technical terms and actions related to mathematical methods and problem-solving

No Comments

No Known Activations