INDEX

Explanations

phrases and terms associated with solutions and treatments in various contexts

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_33/width_131k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.33.hook_resid_post

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

voort

-0.37

 дописавши

-0.36

 nahilalakip

-0.35

DateTimeField

-0.34

 estekak

-0.34

 mergeFrom

-0.33

 useHistory

-0.33

 useNavigate

-0.31

kwe

-0.31

ymce

-0.30

POSITIVE LOGITS

 solution

3.27

solution

2.98

 Solution

2.94

 solutions

2.89

Solution

2.88

 SOLUTION

2.75

 solución

2.59

 solu

2.58

 Solutions

2.55

solutions

2.50

Activations Density 0.055%

phrases and terms associated with solutions and treatments in various contexts

No Comments

No Known Activations