INDEX

Explanations

instances of the word "to."

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

 construcciones

-0.47

 monasterio

-0.46

 nubes

-0.44

 laaj

-0.44

 gebre

-0.42

 finanzas

-0.42

 lazos

-0.42

 blusas

-0.42

 negó

-0.41

 inversiones

-0.40

POSITIVE LOGITS

get

0.70

⟬

0.65

get

0.65

.$_

0.60

TargetException

0.58

 reach

0.57

Get

0.57

access

0.57

touch

0.57

tolo

0.57

Activations Density 0.005%

instances of the word "to."

No Comments

No Known Activations