INDEX

Explanations

phrases indicating causes, effects, or relationships between different concepts

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

 '\\;'

-0.60

сылкі

-0.59

 CURIAM

-0.58



-0.57

 onAnimation

-0.56

 OMITBAD

-0.55

 AssemblyCulture

-0.55

RectangleBorder

-0.54

 Biôgrafia

-0.51

 nonUne

-0.51

POSITIVE LOGITS

Espèce

0.39

 habet

0.38

 كومونز

0.38

 Grecs

0.36

fören

0.35

 Utah

0.35

 convite

0.34

 spéciaux

0.34

 Khusus

0.32

öv

0.32

Activations Density 0.035%

phrases indicating causes, effects, or relationships between different concepts

No Comments

No Known Activations