INDEX

Explanations

references to ordinal numbers, particularly fourth and higher

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

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

ancers

-0.57

Leary

-0.55

digarh

-0.55

alaya

-0.54

UCLA

-0.54

Kolkata

-0.53

 Ahnung

-0.52

 estekak

-0.52

elines

-0.52

PAC

-0.52

POSITIVE LOGITS

 seventh

0.96

 fifth

0.95

 fourth

0.92

 third

0.91

 sixth

0.89

 Seventh

0.88

 ninth

0.86

 FIFTH

0.85

 eighth

0.85

 SEVENTH

0.83

Activations Density 0.012%

references to ordinal numbers, particularly fourth and higher

No Comments

No Known Activations