INDEX

Explanations

sentences that end in "to" followed by high activations

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Configuration

jbloom/Gemma-2b-Residual-Stream-SAEs/gemma_2b_blocks.6.hook_resid_post_16384_anthropic_fast_lr

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

HuggingFaceFW/fineweb

Features

16,384

Data Type

torch.float32

Hook Point

blocks.6.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

<bos>

-2.96

ⓧ

-0.80

/**

-0.77

<?

-0.76

-0.70

الد

-0.63

putnik

-0.59

/*

-0.57

//

-0.57

外

-0.56

POSITIVE LOGITS

 lamborghini

1.54

 scrat

1.53

 affor

1.52

 maneu

1.48

 impra

1.47

 accla

1.41

 disreg

1.39

 panama

1.38

 isuzu

1.37

 Minang

1.36

Activations Density 0.224%

sentences that end in "to" followed by high activations

No Comments

No Known Activations