INDEX

Explanations

computer code, URLs

np_max-act · gemini-2.0-flash

This neuron responds to internal instruction‐format and policy tokens (e.g. “Agent,” “QUEST,” “DATA,” parameter names and underscores) from the model’s system or meta prompt.

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

�

-0.06

_demand

-0.06

 PEOPLE

-0.06

beb

-0.06

curso

-0.06

更新

-0.06

 nozzle

-0.06

 estados

-0.06

]));
↵

-0.05

 vitality

-0.05

POSITIVE LOGITS

ження

0.08

ність

0.07

pluck

0.07

Ret

0.07

/custom

0.07

prm

0.07

ность

0.06

 argparse

0.06

 dalam

0.06

 unravel

0.06

Activations Density 0.012%

computer code, URLs

This neuron responds to internal instruction‐format and policy tokens (e.g. “Agent,” “QUEST,” “DATA,” parameter names and underscores) from the model’s system or meta prompt.

No Comments

No Known Activations