INDEX

Explanations

tokens with specific formatting and structure within a code or data context

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

programming and localization formats

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

 AssemblyCompany

-0.35

 World

-0.34

 user

-0.33

men

-0.33

斓

-0.33

CD

-0.33

BD

-0.32

 utilities

-0.32

IS

-0.32

POSITIVE LOGITS

mybatisplus

0.83

 كومونز

0.73

WithIOException

0.72

EndContext

0.72

IsMutable

0.71

awtextra

0.70

 חיצוניים

0.70

Билгалдахарш

0.67

httphttps

0.65

 Chwiliwch

0.64

Activations Density 0.003%

tokens with specific formatting and structure within a code or data context

programming and localization formats

No Comments

No Known Activations