INDEX

Explanations

terms related to clients and connections in network communication

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

client connection events

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

 مرئيه

-0.36

 AssemblyProduct

-0.35

LookAnd

-0.35

 narrow

-0.33

 výsled

-0.32

vé

-0.32

 Sanford

-0.31

 hjælp

-0.31

ValueStyle

-0.31

Narrow

-0.31

POSITIVE LOGITS

gonic

0.60

 oprot

0.59

httphttps

0.55

 destr

0.48

Terminal

0.47

 Normdatei

0.47

 חיצוניים

0.46

Географиясе

0.46

ſelves

0.45

ClientSize

0.45

Activations Density 0.015%

terms related to clients and connections in network communication

client connection events

No Comments

No Known Activations