INDEX

Explanations

directional indicators and spatial relationships in descriptions

oai_token-act-pair · gpt-4o-mini Triggered by @bot

directions: top, bottom, left, right

np_acts-logits-general · gemini-2.5-flash-lite

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

MeasureSpec

-0.39

RAG

-0.36

}}}{\

-0.36

 createSlice

-0.36

mongoose

-0.35

ैन

-0.32

ioterapia

-0.32

 useful

-0.32

 onFailure

-0.32

 painful

-0.31

POSITIVE LOGITS

 propOrder

0.79

AccessorTable

0.79

 initComponents

0.73

 Theſe

0.73

featureID

0.67

 myſelf

0.66

Datuak

0.62

Efq

0.60

 themſelves

0.59

 houſe

0.59

Activations Density 0.009%

directional indicators and spatial relationships in descriptions

directions: top, bottom, left, right

No Comments

No Known Activations