INDEX

Explanations

punctuation marks and indicators of discourse structure

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

RefNanny

-0.54

withIdentifier

-0.52

 AssemblyTitle

-0.48

 мәкал

-0.48

kháu

-0.48

]<<"

-0.47

 nakalista

-0.46

ftagPool

-0.45

EndContext

-0.44

 ffilmiau

-0.44

POSITIVE LOGITS

 jScrollPane

0.36

ьаж

0.35

 vare

0.35

AutoScaleMode

0.35

เลย

0.33

 aeroporto

0.32

 correctes

0.32

 Ries

0.32

Ries

0.32

 https

0.32

Activations Density 0.260%

punctuation marks and indicators of discourse structure

No Comments

No Known Activations