INDEX

Explanations

terms related to time, particularly focusing on the concept of 'year' and 'round' patterns

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

 Söhne

-0.41

 Jaffe

-0.41

 mijne

-0.40

 Aufs

-0.40

 Sünden

-0.37

 Fahr

-0.36

énario

-0.36

 Beim

-0.35

 infeliz

-0.35

 terceira

-0.35

POSITIVE LOGITS

AnchorStyles

0.69

ſelf

0.69

 purpoſe

0.68

 MainAxisSize

0.67

ſelves

0.65

 ſtand

0.62

 seasonally

0.61

 fubject

0.60

Summer

0.60

 Houſe

0.59

Activations Density 0.007%

terms related to time, particularly focusing on the concept of 'year' and 'round' patterns

No Comments

No Known Activations