INDEX

Explanations

instances of the term "DOM" or related variations in document contexts

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

withIdentifier

-0.77

Parcelize

-0.71

úrate

-0.63

ướ

-0.61

lerc

-0.61

 obligé

-0.60

 japones

-0.58

 Apalagi

-0.58

 Murillo

-0.58

────────

-0.58

POSITIVE LOGITS

dom

1.23

DOM

1.21

Dom

1.11

dom

1.10

Dom

1.04

 Dominguez

1.02

doms

0.93

 Domes

0.93

DOM

0.91

 Dames

0.80

Activations Density 0.006%

instances of the term "DOM" or related variations in document contexts

No Comments

No Known Activations