INDEX

Explanations

terms associated with legal and medical contexts, especially focusing on successful outcomes and procedural elements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.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

ніципалі

-0.66

SequentialGroup

-0.60

 Meksiku

-0.57

 незавершена

-0.55

 שוליים

-0.52

 confronti

-0.51

 ***!

-0.50

AutoScaleMode

-0.50

Geplaatst

-0.50

تقاوى

-0.49

POSITIVE LOGITS

 SafeMath

0.46

 saites

0.33

⮕

0.33

wa

0.32

voll

0.32

Rel

0.32

 Rite

0.32

 mete

0.32

あた

0.32

 micro

0.32

Activations Density 0.058%

terms associated with legal and medical contexts, especially focusing on successful outcomes and procedural elements

No Comments

No Known Activations