INDEX

Explanations

references to awards and their associated descriptions

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_33/width_131k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.33.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

 հղումներ

-0.48

principalColumn

-0.47

 ब्रेकडाउन

-0.45

хьтан

-0.44

 PeEnEo

-0.43

 Wikimedijinoj

-0.41

">//

-0.41

 Мексичка

-0.41

ługość

-0.39

 Roskov

-0.39

POSITIVE LOGITS

 awarded

0.85

 award

0.72

 Awarded

0.71

 awarding

0.69

 awards

0.67

 bestowed

0.66

 recognising

0.65

 recognizing

0.65

 recognizes

0.65

 annually

0.64

Activations Density 0.014%

references to awards and their associated descriptions

No Comments

No Known Activations