INDEX

Explanations

references to numerous entities or large quantities in various contexts

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

specifying large numbers

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



-0.49

 astore

-0.48

 transaction

-0.43

enc

-0.43

SequentialGroup

-0.41

fillRect

-0.40

 Qualified

-0.40

UMS

-0.39

PushButton

-0.39

 لط

-0.39

POSITIVE LOGITS

featureID

0.66

 ribuan

0.48

 thousands

0.47

thousands

0.47

 hundreds

0.46

hundreds

0.46

 ratusan

0.45

Tikang

0.45

ousands

0.44

 numerus

0.44

Activations Density 0.069%

references to numerous entities or large quantities in various contexts

specifying large numbers

No Comments

No Known Activations