INDEX

Explanations

elements related to media and news programming formats

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

interview shows and programs

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

webElementXpaths

-0.78

oa̍t

-0.68

 nakalista

-0.67

CloseOperation

-0.63

 onCreateView

-0.61

steamcommunity

-0.60

 oprot

-0.59

TestingModule

-0.57

AccessorTable

-0.56

enablog

-0.56

POSITIVE LOGITS

 interview

0.35

 interviews

0.32

 stock

0.29

 guests

0.29

 show

0.29

 interviewing

0.29

 search

0.27

scrollTo

0.26

 camer

0.26

 script

0.26

Activations Density 0.026%

elements related to media and news programming formats

interview shows and programs

No Comments

No Known Activations