INDEX

Explanations

phrases related to evaluation and accuracy in performance descriptions

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

accurately reconstructed

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

PerSecond

-0.35

 kafe

-0.34

apo

-0.34

 Medical

-0.33

 Academic

-0.32

Apo

-0.32

)_/¯

-0.31

Outbound

-0.31

 feen

-0.30

lebnis

-0.30

POSITIVE LOGITS

 للاسماء

0.67

rungsseite

0.66

 Meksiku

0.60

 EconPapers

0.56

ſchaft

0.56

AnchorStyles

0.56

 HasFactory

0.54

 мәкал

0.53

 المعيارى

0.52

endpush

0.52

Activations Density 0.120%

phrases related to evaluation and accuracy in performance descriptions

accurately reconstructed

No Comments

No Known Activations