INDEX

Explanations

how to

np_max-act · gemini-2.0-flash

The neuron fires on words that cue explanations or instructions—tokens like “how,” “instructions,” and similar explanatory markers.

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ddit

-0.07

wo

-0.07

 Barbie

-0.06

Rod

-0.06

 carved

-0.06

.Code

-0.06

ldata

-0.06

(Value

-0.06

 nepří

-0.06

docs

-0.06

POSITIVE LOGITS

 totalitarian

0.07

ंघ

0.07

Upper

0.07

골

0.07

 pharmaceutical

0.07

 Environmental

0.07

 filling

0.06

 sulfate

0.06

ัป

0.06

 khuôn

0.06

Activations Density 0.036%

how to

The neuron fires on words that cue explanations or instructions—tokens like “how,” “instructions,” and similar explanatory markers.

No Comments

No Known Activations