INDEX

Explanations

stories

np_max-act · gemini-2.0-flash

This neuron detects the special tokens used to mark header or metadata boundaries in the conversation (e.g. the “<|start_header_id|>” and “<|end_header_id|>” 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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

"?

-0.07

.wh

-0.06

 Simulation

-0.06

_WRONLY

-0.06

 {}".

-0.06

uras

-0.06

์ฟ

-0.06

_CNT

-0.06

７

-0.06

POSITIVE LOGITS

 deficiency

0.07

dbName

0.07

ACK

0.06

 recycled

0.06

 král

0.06

 acknowledge

0.06

grund

0.06

	events

0.06

 ControllerBase

0.06

 closures

0.06

Activations Density 0.009%

stories

This neuron detects the special tokens used to mark header or metadata boundaries in the conversation (e.g. the “<|start_header_id|>” and “<|end_header_id|>” markers).

No Comments

No Known Activations