INDEX

Explanations

points of emphasis or significant breaks in the narrative

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

gänger

-0.40

 risa

-0.35

 redor

-0.34

 surn

-0.32

 dayan

-0.32

הערות

-0.31

cienza

-0.31

новниш

-0.31

 Espec

-0.30

say

-0.30

POSITIVE LOGITS

-------

0.57

-/**

0.57

0.55

 firebaseConfig

0.55

 uLocal

0.54

éroport

0.51

dAtA

0.50

modelBuilder

0.49

//{

0.49

 ModelRenderer

0.49

Activations Density 0.126%

points of emphasis or significant breaks in the narrative

No Comments

No Known Activations