INDEX

Explanations

closing braces and related syntax characters in code

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

consequent loss

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

UserScript

-0.77

 betweenstory

-0.77

PerformLayout

-0.74

杓

-0.70

enderror

-0.69

 MainAxisSize

-0.69

 queſta

-0.69

 transfieras

-0.68

tangentMode

-0.68

windowFixed

-0.68

POSITIVE LOGITS

 Umum

0.53

 Grecs

0.44

 Gnade

0.43

 antidesliz

0.43

 Púb

0.43

 Prancis

0.41

 Freundschaft

0.40

 Usaha

0.40

 Jenderal

0.40

 Vainqueur

0.39

Activations Density 0.001%

closing braces and related syntax characters in code

consequent loss

No Comments

No Known Activations