INDEX

Explanations

references to scientific studies, findings, and methodologies

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

Adverbial modifiers indicating degree or extent, often used to qualify assertions by showing how complete or partial a relationship or causation is.

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

TIC

-0.52

 ск

-0.47

SSN

-0.46

 Honor

-0.46

偶

-0.45

VID

-0.44

شنبه

-0.44

NBC

-0.44

lamine

-0.44

NV

-0.44

POSITIVE LOGITS

mostly

0.56

 mostly

0.56

 głównie

0.56

 chiefly

0.54

mainly

0.53

 mainly

0.52

Mostly

0.52

WebVitals

0.50

ほぼ

0.50

 principalement

0.49

Activations Density 0.170%

references to scientific studies, findings, and methodologies

Adverbial modifiers indicating degree or extent, often used to qualify assertions by showing how complete or partial a relationship or causation is.

No Comments

No Known Activations