INDEX

Explanations

specific names, terms, or entities related to diverse subjects including language, names of places, and personal communication

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

tokens followed by specific suffixes

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

참고

-0.35

 harde

-0.31

 spesa

-0.30

疡

-0.30

stande

-0.30

Problem

-0.30

 Mala

-0.30

 problem

-0.29

 blink

-0.29

 timely

-0.29

POSITIVE LOGITS

########.

0.73

 Biôgrafia

0.63

ьаж

0.60

 nonUne

0.60

ỡng

0.59

Personensuche

0.59

 فريبيس

0.58

fromnode

0.57

estacks

0.56

፩

0.55

Activations Density 0.117%

specific names, terms, or entities related to diverse subjects including language, names of places, and personal communication

tokens followed by specific suffixes

No Comments

No Known Activations