INDEX

Explanations

specific locations, especially geographic or building names

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

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

ngdoc

-0.39

Personendaten

-0.39

分明

-0.33

 Upper

-0.33

 opge

-0.33

 plenamente

-0.31

 iprot

-0.31

 Imperio

-0.31

 feiert

-0.31

 linh

-0.30

POSITIVE LOGITS

 ComVisible

0.59

PerformLayout

0.54

 صوتيه

0.53

UrlResolution

0.53

 initComponents

0.52

 disambiguazione

0.51

jsxFileName

0.49

 <=",

0.49

ніципалі

0.49

 preſent

0.49

Activations Density 0.072%

specific locations, especially geographic or building names

No Comments

No Known Activations