INDEX

Explanations

technical code structures and configurations

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_23/width_131k/average_l0_101

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.23.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

PYX

-0.63

 ivelany

-0.60

 ModelExpression

-0.59

 ***!

-0.57

saraba

-0.56

@"/

-0.55

 Vikipedi

-0.53

новништво

-0.53

RenderAtEndOf

-0.52

parsedMessage

-0.51

POSITIVE LOGITS

AutoScaleMode

0.38

 which

0.38

 koszulka

0.37

 Werkzeug

0.36

 abond

0.36

 istället

0.36

 siquiera

0.36

 nonUne

0.35

 pungkasnya

0.35

indest

0.35

Activations Density 0.155%

technical code structures and configurations

No Comments

No Known Activations