INDEX

Explanations

responses

np_max-act · gemini-2.0-flash

This neuron activates on instruction tokens that prompt the model to explain or justify its answer (e.g., words like “reasoning,” “answer,” “explain,” “your”).

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

 monsters

-0.06

 resizable

-0.06

Plant

-0.06

 мам

-0.06

оит

-0.06

NotSupportedException

-0.06

 hodnoty

-0.06

║

-0.06

Wer

-0.06

 LORD

-0.06

POSITIVE LOGITS

-da

0.07

_MATH

0.07

Sai

0.07

/kg

0.07

-state

0.06

 vows

0.06

-US

0.06

 суще

0.06

ظٹط

0.06

 yuan

0.06

Activations Density 0.013%

responses

This neuron activates on instruction tokens that prompt the model to explain or justify its answer (e.g., words like “reasoning,” “answer,” “explain,” “your”).

No Comments

No Known Activations

responses

This neuron activates on instruction tokens that prompt the model to explain or justify its answer (e.g., words like “reasoning,” “answer,” “explain,” “your”).

No Comments

No Known Activations