INDEX

Explanations

Code and technical documents

np_max-act · gemini-2.0-flash

This neuron activates on the prompt’s tool‐usage/instruction headings—particularly keywords that introduce or describe how to format tool calls (e.g. “Search,” “Input,” “query,” “should,” etc.).

oai_token-act-pair · o4-mini Triggered by @xinyanhu8

instructions for retrieving information from an online source.

oai_token-act-pair · gpt-4o-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

ffmpeg

-0.06

 таб

-0.06

佐

-0.06

-links

-0.06

prak

-0.06

>w

-0.06

 tick

-0.06

.cross

-0.06

ประจำ

-0.06

.endpoint

-0.06

POSITIVE LOGITS

 Vegetable

0.07

 desn

0.07

 Activate

0.06

로그

0.06

 ALSO

0.06

frau

0.06

 wondering

0.06

 Pick

0.06

раль

0.06

 commemor

0.06

Activations Density 0.010%

Code and technical documents

This neuron activates on the prompt’s tool‐usage/instruction headings—particularly keywords that introduce or describe how to format tool calls (e.g. “Search,” “Input,” “query,” “should,” etc.).

instructions for retrieving information from an online source.

No Comments

No Known Activations

Code and technical documents

This neuron activates on the prompt’s tool‐usage/instruction headings—particularly keywords that introduce or describe how to format tool calls (e.g. “Search,” “Input,” “query,” “should,” etc.).

instructions for retrieving information from an online source.

No Comments

No Known Activations