INDEX

Explanations

phrases indicating assistance or support

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

The neuron activates for the word "help" when it appears in contexts describing assistance, support, or beneficial actions (typically in phrases like "will help" or "can help").

oai_token-act-pair · claude-4-5-sonnet Triggered by @jyhe0408

phrases asserting that something provides assistance to someone in achieving a goal or facilitating an outcome.

oai_token-act-pair · gpt-5 Triggered by @jyhe0408

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 15-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L15R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.15.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

iya

-0.16

à¸«à¸¡à¸²à¸¢

-0.15

oled

-0.15

lou

-0.15

ãģ¹ãģį

-0.15

olest

-0.15

naments

-0.15

ize

-0.15

cribe

-0.15

rak

-0.15

POSITIVE LOGITS

us

0.24

desk

0.20

fully

0.19

 Äĳá»¡

0.19

me

0.19

lessly

0.17

esch

0.16

you

0.15

ERSHEY

0.14

Activations Density 0.061%

phrases indicating assistance or support

The neuron activates for the word "help" when it appears in contexts describing assistance, support, or beneficial actions (typically in phrases like "will help" or "can help").

phrases asserting that something provides assistance to someone in achieving a goal or facilitating an outcome.

No Comments

No Known Activations