INDEX

Explanations

phrases relating to the quantity and frequency of publications in research contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 17-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_17/width_65k/average_l0_125

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.17.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

การ

-0.52

gemeente

-0.49

titleMargin

-0.49

}`).

-0.48

<bos>

-0.48

 recientes

-0.48

libft

-0.47

quiao

-0.46

tado

-0.46

 }}">

-0.46

POSITIVE LOGITS

 myſelf

0.82

 nahilalakip

0.76

 themſelves

0.75

 Majefty

0.74

 itſelf

0.73

 himſelf

0.71

 ſeveral

0.71

Diwedd

0.66

 ſever

0.65

 '\\;'

0.65

Activations Density 0.276%

phrases relating to the quantity and frequency of publications in research contexts

No Comments

No Known Activations