INDEX

Explanations

references to geometric or physical axes and their coordinates

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

füllung

-0.48

 convinced

-0.44

 Mayr

-0.43

 stamped

-0.43

PCell

-0.42

WriteTagHelper

-0.42

ran

-0.41

 admitted

-0.41

tanleria

-0.40

ismet

-0.40

POSITIVE LOGITS

 skolen

0.47

awtextra

0.47

rungsseite

0.47

 kağı

0.46

 Italijanski

0.44

 FetchType

0.44

 problemer

0.44

 zoude

0.43

AndroidJUnit

0.43

 språk

0.43

Activations Density 0.036%

references to geometric or physical axes and their coordinates

No Comments

No Known Activations