INDEX

Explanations

references to specific numeric values or identifiers in a programming or technical context

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

referential

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

ppuden

-0.45

ácií

-0.42

malige

-0.41

 dodatk

-0.41

URLException

-0.40

 mohair

-0.40

 taupe

-0.40

 Economía

-0.40

 gider

-0.40

 clogs

-0.39

POSITIVE LOGITS

 Referential

3.75

 refer

1.48

 Refer

1.41

Refer

1.37

refer

1.31

 REFER

1.27

REFER

1.12

 Reference

1.06

 reference

1.03

 referring

1.03

Activations Density 0.001%

references to specific numeric values or identifiers in a programming or technical context

referential

No Comments

No Known Activations