INDEX

Explanations

connections to known elements or established facts in various contexts

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

Technical or specific descriptors in academic/scientific text that refer to parameters, variables, properties, or items being defined, measured, or referenced in their domain context.

eleuther_acts_top20 · claude-3-7-sonnet-20250219 Triggered by @lu-christina

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

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

 ſche

-0.72

httphttps

-0.68

 ſch

-0.64

 surla

-0.60

 ſte

-0.60

 Perſ

-0.60

 viſ

-0.58

 ſol

-0.58

 ſever

-0.56

ArgsConstructor

-0.55

POSITIVE LOGITS

 Ahnung

0.33

 Abgerufen

0.32

 Rechnung

0.32

 ویکی‌پدیای

0.31

 hjemmes

0.30

 Ehre

0.29

板

0.29

世

0.29

 honneur

0.29

 Lightboxes

0.29

Activations Density 0.237%

connections to known elements or established facts in various contexts

Technical or specific descriptors in academic/scientific text that refer to parameters, variables, properties, or items being defined, measured, or referenced in their domain context.

No Comments

No Known Activations