INDEX

Explanations

instances of text that indicate the start of a new section or notable emphasis

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

New Auto-Interp

Configuration

jbloom/Gemma-2b-Residual-Stream-SAEs/gemma_2b_blocks.12.hook_resid_post_16384

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

HuggingFaceFW/fineweb

Features

16,384

Data Type

torch.float32

Hook Point

blocks.12.hook_resid_post

Architecture

standard

Context Size

1,024

Dataset

HuggingFaceFW/fineweb

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 massacres

-0.84

 despotism

-0.84

 traitors

-0.80

 Fascism

-0.74

 treachery

-0.74

 armaments

-0.74

 belliger

-0.73

 disinformation

-0.73

 demoral

-0.72

 anarchy

-0.69

POSITIVE LOGITS

<bos>

14.27

expandindo

2.29

Autoritní

2.27

GEBURTSDATUM

2.22

 Administrativna

2.18

 betweenstory

2.11

تقاوى

2.10

 Italijani

1.91

 kasarigan

1.90

 дописавши

1.89

Activations Density 0.918%

instances of text that indicate the start of a new section or notable emphasis

No Comments

No Known Activations