INDEX

Explanations

try

np_max-act · gemini-2.0-flash

The neuron detects “try to …” constructions—i.e. when the assistant tentatively offers to perform an action.

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

ある

-0.06

 department

-0.06

 اندازه

-0.06

 replicate

-0.06

_AD

-0.06

出した

-0.06

 websites

-0.06

PARTMENT

-0.06

 expansive

-0.06

 resurgence

-0.06

POSITIVE LOGITS

Converter

0.08

 проф

0.07

(ErrorMessage

0.06

_trial

0.06

سن

0.06

将

0.06

[attr

0.06

IBAction

0.06

.Dispatch

0.06

 verze

0.06

Activations Density 0.013%

try

The neuron detects “try to …” constructions—i.e. when the assistant tentatively offers to perform an action.

No Comments

No Known Activations