INDEX

Explanations

right and wrong

np_max-act · gemini-2.0-flash

The neuron activates on words and phrases expressing truth, correctness, or their opposites (e.g. “true,” “truth,” “wrong”).

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

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

 товар

-0.06

.parse

-0.06

 Когда

-0.06

 Gaussian

-0.06

 Hungarian

-0.06

 sectors

-0.06

१�

-0.06

homepage

-0.06

 invention

-0.06

 fireplace

-0.06

POSITIVE LOGITS

 hitch

0.08

Route

0.07

 Claim

0.07

.getConfig

0.07

 cautiously

0.07

_MIN

0.07

ậc

0.07

 thats

0.06

ilit

0.06

 Tiny

0.06

Activations Density 0.127%

right and wrong

The neuron activates on words and phrases expressing truth, correctness, or their opposites (e.g. “true,” “truth,” “wrong”).

No Comments

No Known Activations