INDEX

Explanations

expressions related to significance or updates in information

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

rungsseite

-0.35

 snippetHide

-0.35

DebuggerStep

-0.34

 typical

-0.32

 richtigen

-0.31

ингл

-0.29

 preguntó

-0.29

 quisiera

-0.29

 stanga

-0.29

 desee

-0.29

POSITIVE LOGITS

 betweenstory

0.61

SequentialGroup

0.59

ftagPool

0.55

endregion

0.53

recent

0.51

suspend

0.50

tvguidetime

0.50



0.50

 témoig

0.49

ConstraintMaker

0.48

Activations Density 0.020%

expressions related to significance or updates in information

No Comments

No Known Activations