INDEX

Explanations

programming constructs and logical conditions

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

github issues

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

uxxxx

-0.54

intios

-0.33

 håll

-0.31

tę

-0.29

 editText

-0.29

 encar

-0.29

 scrap

-0.28

SEP

-0.27

데

-0.26

 cremallera

-0.26

POSITIVE LOGITS

MigrationBuilder

0.90

PreferredItem

0.77

bootstrapcdn

0.71

 utafitiHapana

0.71

 typelib

0.70

complexContent

0.68

ConstraintMaker

0.67

 AssemblyCompany

0.64

NamedQueries

0.64

ValueGenerated

0.63

Activations Density 0.259%

programming constructs and logical conditions

github issues

No Comments

No Known Activations