INDEX

Explanations

terms related to sequences and classifications in various contexts

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

determiner noun

np_acts-logits-general · gemini-2.5-flash-lite

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

-0.38

 because

-0.36

 instead

-0.36

들

-0.34

 како

-0.33

NameInMap

-0.33

 mivel

-0.32

SceneManagement

-0.32

 anstatt

-0.32

POSITIVE LOGITS

 beſch

0.69

iſen

0.62

 verſch

0.59

 ſta

0.59

 minOccurs

0.58

ðsíða

0.58

 myſelf

0.58

 unſer

0.57

 quelqu

0.56

 ſeine

0.56

Activations Density 0.200%

terms related to sequences and classifications in various contexts

determiner noun

No Comments

No Known Activations