INDEX

Explanations

specific identifiers and numerical values related to versions and sequences in software or biological contexts

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

PCV3

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

ThroughAttribute

-0.48

 AssemblyTitle

-0.42

AdapterView

-0.41

 autorytatywna

-0.41

 http

-0.40

imageshack

-0.39

writeFieldEnd

-0.38

:][

-0.38

 cref

-0.36

textLabel

-0.35

POSITIVE LOGITS

 myſelf

0.66

 snippetHide

0.65

令和

0.65

 Obrador

0.64

 ſtand

0.63

 itſelf

0.62

dflare

0.59

SBATCH

0.59

 NVMe

0.59

 étoit

0.59

Activations Density 0.025%

specific identifiers and numerical values related to versions and sequences in software or biological contexts

PCV3

No Comments

No Known Activations