INDEX

Explanations

numerical patterns, specifically sequences involving numbers and abbreviations

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

<bos>

-3.23

ⓧ

-1.08

Autoritní

-0.79

-0.75

/**

-0.72

 inaugurate

-0.68

Портали

-0.68

///**

-0.67

/***

-0.67

contentLoaded

-0.67

POSITIVE LOGITS

 affor

1.11

 Intere

1.01

 increa

1.00

 maneu

1.00

 pollut

0.97

 bandung

0.96

 pleins

0.96

 véhic

0.95

 stockholm

0.92

 maroc

0.92

Activations Density 0.374%

numerical patterns, specifically sequences involving numbers and abbreviations

No Comments

No Known Activations