INDEX

Explanations

Structured text/code

np_max-act · gemini-2.0-flash

sentences that describe mindfulness and decision-making.

oai_token-act-pair · gpt-4o-mini Triggered by @xinyanhu8

This neuron responds to the model’s internal control and formatting tokens (e.g. <|start_header_id|>, <|end_header_id|>, <|eot_id|>), marking message boundaries and roles.

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

-0.07

とする

-0.07

_iteration

-0.07

respond

-0.06

(lhs

-0.06

Ž

-0.06

creation

-0.06

})

-0.06

 Vill

-0.06

^^

-0.06

POSITIVE LOGITS

NTAX

0.07

 Stopwatch

0.06

 bordel

0.06

 Costco

0.06

잔

0.06

.parentElement

0.06

 Persons

0.06

\Contracts

0.06

PECT

0.06

 mực

0.06

Activations Density 0.037%

Structured text/code

sentences that describe mindfulness and decision-making.

This neuron responds to the model’s internal control and formatting tokens (e.g. <|start_header_id|>, <|end_header_id|>, <|eot_id|>), marking message boundaries and roles.

No Comments

No Known Activations