INDEX

Explanations

references to "Star Wars" and related terminology

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

we

-1.19

our

-1.06

We

-0.98

We

-0.91

 ourselves

-0.88

我們

-0.85

我们

-0.84

Our

-0.78

我们就

-0.77

our

-0.76

POSITIVE LOGITS

NOPQRST

0.68

 Plaintiff

0.52

 Encyclopaedia

0.52

 tunik

0.50

 Belinda

0.49

 Distinction

0.49

 Catto

0.48

ритори

0.47

 MERCHANTABILITY

0.47

thodoxy

0.47

Activations Density 0.054%

references to "Star Wars" and related terminology

No Comments

No Known Activations