INDEX

Explanations

programming-related syntax, particularly in the context of code and metadata attributes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 18-gemmascope-mlp-65k

Configuration

google/gemma-scope-2b-pt-mlp/layer_18/width_65k/average_l0_88

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.18.hook_mlp_out

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

 fire

-0.54

 wildfires

-0.53

me

-0.50

UIControlState

-0.50

rift

-0.48

 torch

-0.48

 flame

-0.47

RegressionTest

-0.46

 fuego

-0.46

vana

-0.45

POSITIVE LOGITS

Glej

0.71

 EconPapers

0.69

ResponseWriter

0.63

 Monfieur

0.62

érité

0.62

NewUrlParser

0.61

GenerationType

0.60

addObject

0.58

 defaultstate

0.57

Efq

0.57

Activations Density 0.010%

programming-related syntax, particularly in the context of code and metadata attributes

No Comments

No Known Activations