INDEX

Explanations

quantitative measures of accuracy and tolerance in statistical or experimental data

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

size and comparison

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

Rptr

-0.47

arashtra

-0.45

 Sagar

-0.44

éroport

-0.43

 PeEnEo

-0.43

 Mérida

-0.42

support

-0.42

gbaar

-0.41

onicus

-0.41

transQ

-0.41

POSITIVE LOGITS

 higher

0.44

 poor

0.43

ThroughAttribute

0.43

expandindo

0.41

 narrow

0.41

 high

0.40

 ModelExpression

0.40

low

0.39

HtmlAttribute

0.39

oa̍t

0.39

Activations Density 0.311%

quantitative measures of accuracy and tolerance in statistical or experimental data

size and comparison

No Comments

No Known Activations