INDEX

Explanations

programming-related types and structures

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

start of turn user

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

 мәкал

-0.93

StoryboardSegue

-0.92

WriteTagHelper

-0.88

 Chwiliwch

-0.86

 betweenstory

-0.86

transQ

-0.82

+#+#

-0.81

TagMode

-0.81

AddHtmlAttribute

-0.79

 дописавши

-0.78

POSITIVE LOGITS

0.29

0.28

 fédé

0.26

 contemporaine

0.25

 Profesional

0.25

 further

0.24

ค์

0.23

Activations Density 0.018%

programming-related types and structures

start of turn user

No Comments

No Known Activations