INDEX

Explanations

Two options or possibilities

np_max-act · gemini-2.0-flash

The neuron activates on the preposition “through,” flagging mentions of processes or methods (e.g. “through clinical trials,” “through post‐marketing surveillance”).

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

Descriptions

-0.06

 groundwork

-0.06

_LANG

-0.06

ircles

-0.06

 patience

-0.06

_parse

-0.06

 pró

-0.06

_world

-0.06

upid

-0.06

_pool

-0.06

POSITIVE LOGITS

 Chỉ

0.07

 κοι

0.07

and

0.07

(select

0.07

banana

0.07

Ket

0.07

oggle

0.07

kot

0.07

 ever

0.07

beb

0.06

Activations Density 0.103%

Two options or possibilities

The neuron activates on the preposition “through,” flagging mentions of processes or methods (e.g. “through clinical trials,” “through post‐marketing surveillance”).

No Comments

No Known Activations