INDEX

Explanations

Code/text errors

np_max-act · gemini-2.0-flash

The neuron is triggered by tokens that begin with the letter “a,” especially common prefixes like “ab-,” “ac-,” “ad-,” etc.

oai_token-act-pair · o4-mini Triggered by @xinyanhu8

It detects subword tokens that are prefixes or beginnings of words starting with the letter "a" (common "a"-word stems like ab-, ac-, ad-, al-).

oai_token-act-pair · gpt-5-mini Triggered by @vetterc0

New Auto-Interp

Configuration

andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1

Dataset (Dashboard)

Various

Features

131,072

Data Type

float32

Hook Name

blocks.11.hook_resid_post

Architecture

standard

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

icast

-0.08

 increasingly

-0.07

 достаточно

-0.07

lst

-0.07

 Witt

-0.06

 persuasive

-0.06

\Database

-0.06

，如

-0.06

如

-0.06

".

-0.06

POSITIVE LOGITS

Tay

0.07

-duration

0.07

ág

0.06

 레이

0.06

_contin

0.06

 цик

0.06

kelig

0.06

Jose

0.06

ATH

0.06

引用频次

0.06

Activations Density 0.822%

Code/text errors

The neuron is triggered by tokens that begin with the letter “a,” especially common prefixes like “ab-,” “ac-,” “ad-,” etc.

It detects subword tokens that are prefixes or beginnings of words starting with the letter "a" (common "a"-word stems like ab-, ac-, ad-, al-).

No Comments

No Known Activations

Code/text errors

The neuron is triggered by tokens that begin with the letter “a,” especially common prefixes like “ab-,” “ac-,” “ad-,” etc.

It detects subword tokens that are prefixes or beginnings of words starting with the letter "a" (common "a"-word stems like ab-, ac-, ad-, al-).

No Comments

No Known Activations