INDEX

Explanations

terms indicating sequence or order, particularly related to the first and last items in a series

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-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

 Tatsache

-0.52

เพราะ

-0.47

 księ

-0.45

ไว้

-0.45

 faptul

-0.44

 världen

-0.44

 sitä

-0.44

 siinä

-0.44

Personensuche

-0.44

 vuotta

-0.43

POSITIVE LOGITS

two

0.63

 third

0.61

few

0.58

 successive

0.58

 second

0.57

 fifth

0.57

 unsuccessful

0.53

interopRequire

0.52

 three

0.52

 several

0.52

Activations Density 0.310%

terms indicating sequence or order, particularly related to the first and last items in a series

No Comments

No Known Activations