INDEX

Explanations

terms related to parsing and processing data

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 1-gemmascope-mlp-16k

Configuration

google/gemma-scope-2b-pt-mlp/layer_1/width_16k/average_l0_105

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.1.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

parse

-2.22

 parse

-2.00

Parse

-1.72

 Parse

-1.65

 parsing

-1.60

 parsed

-1.48

parsed

-1.27

Parsing

-1.19

parsing

-1.18

PARSE

-1.17

POSITIVE LOGITS

 OMITBAD

0.63

 initComponents

0.56

łucha

0.55

таратура

0.53

өз

0.52

aperto

0.52

onAttach

0.49

 spie

0.49

irritation

0.48

 invokingState

0.48

Activations Density 0.139%

terms related to parsing and processing data

No Comments

No Known Activations