INDEX

Explanations

specific formatting or structure in the text, likely headers or document sections

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

Jeografia

-0.75

 nonUne

-0.73

INSEE

-0.65

鄖

-0.61

🢐

-0.61

 strdup

-0.60

⤹

-0.60

 DbSet

-0.60

*)&

-0.57

ptonshire

-0.57

POSITIVE LOGITS

LEGGI

0.47

 órg

0.41

 ensaio

0.41

 abenço

0.40

 måde

0.39

 avoient

0.37

 hayas

0.36

 pouvoit

0.36

intéress

0.36

 feroit

0.35

Activations Density 0.000%

specific formatting or structure in the text, likely headers or document sections

No Comments

No Known Activations