INDEX

Explanations

clauses indicating positive or negative sentiments

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

WriteString

-0.49

jgl

-0.46

MigrationBuilder

-0.45

 propOrder

-0.44

CIR

-0.42

 mergeFrom

-0.41

 noDo

-0.40

nomin

-0.40

CIR

-0.40

Pick

-0.40

POSITIVE LOGITS

 others

0.53

Others

0.52

 Others

0.51

 sommige

0.50

 antaranya

0.50

hæng

0.49

 graça

0.49

latego

0.49

中には

0.48

 popoli

0.48

Activations Density 0.093%

clauses indicating positive or negative sentiments

No Comments

No Known Activations