INDEX

Explanations

coding algorithms

np_max-act · gemini-2.0-flash

data structures used in the solution code.

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

The neuron fires on tokens that are part of algorithm‐problem identifiers or function names (e.g. “Duplicate,” “Consecutive,” “Matrix,” “Combinations,” “Zero,” etc.).

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

 리스트

-0.07

 Copyright

-0.06

JSONArray

-0.06

 визнач

-0.06

 حتی

-0.06

 删除

-0.06

_CATEGORY

-0.06

ुर

-0.06

WhiteSpace

-0.06

 Dialogue

-0.06

POSITIVE LOGITS

.Progress

0.07

mah

0.06

actor

0.06

	state

0.06

 interoper

0.06

.Symbol

0.06

Mari

0.06

.plan

0.05

inker

0.05

.HtmlControls

0.05

Activations Density 0.010%

coding algorithms

data structures used in the solution code.

The neuron fires on tokens that are part of algorithm‐problem identifiers or function names (e.g. “Duplicate,” “Consecutive,” “Matrix,” “Combinations,” “Zero,” etc.).

No Comments

No Known Activations