INDEX

Explanations

dialogue and conversational phrases

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

declarations of problems

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

 quedado

-0.39

Correct

-0.37

 quedará

-0.37

 saludos

-0.35

 correct

-0.35

เยอะ

-0.35

 jamás

-0.35

 quedan

-0.35

 précisément

-0.35

 fácilmente

-0.34

POSITIVE LOGITS

RenderAtEndOf

0.73

хьтан

0.67

 betweenstory

0.65

verifyException

0.64

AndEndTag

0.64

UnknownFieldSet

0.61

 noDo

0.60

UrlResolution

0.60

 AssemblyCompany

0.60

LabelTagHelper

0.60

Activations Density 0.045%

dialogue and conversational phrases

declarations of problems

No Comments

No Known Activations