Dataset Preview
Duplicate
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
[{'expected': SplitInfo(name='train', num_bytes=125845737, num_examples=50000, shard_lengths=None, original_shard_lengths=None, dataset_name=None), 'recorded': SplitInfo(name='train', num_bytes=26019128, num_examples=1219, shard_lengths=None, original_shard_lengths=None, dataset_name='cygnis-alpha2-instruct-mix')}]
Error code:   UnexpectedError

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

instruction
string
context
string
response
string
source
string
text
string
Generate an executable Python function generated from the given prompt. The function should take stdin as input and print the output. Simply call the function after the definition.You are given a binary matrix A of size n × n. Let's denote an x-compression of the given matrix as a matrix B of size n/x × n/x such that f...
Expertise Code - Souveraineté Cygnis
[RÉFLEXION] Analyse approfondie de la problématique code. Validation : Protocole de haute précision Simonc-44 activé. [RÉPONSE] <|begin_of_thought|> Okay, let's see. The problem is to find the maximum x such that the given binary matrix can be x-compressed. An x-compression means that the matrix is divided into x by...
null
null
Return your final response within \boxed{}. Michael writes down all the integers between 1 and \( N \) inclusive on a piece of paper and discovers that exactly \( 40\% \) of them have leftmost digit 1. Given that \( N > 2017 \), find the smallest possible value of \( N \).
Expertise Logic - Souveraineté Cygnis
[RÉFLEXION] Analyse approfondie de la problématique logic. Validation : Protocole de haute précision Simonc-44 activé. [RÉPONSE] <|begin_of_thought|> Okay, so Michael wrote down all the integers from 1 to N, and found that exactly 40% of them have a leftmost digit of 1. We need to find the smallest possible N that's...
null
null
Generate an executable Python function generated from the given prompt. The function should take stdin as input and print the output. Simply call the function after the definition.Andrew and Jerry are playing a game with Harry as the scorekeeper. The game consists of three rounds. In each round, Andrew and Jerry draw r...
Expertise Code - Souveraineté Cygnis
[RÉFLEXION] Analyse approfondie de la problématique code. Validation : Protocole de haute précision Simonc-44 activé. [RÉPONSE] <|begin_of_thought|> Okay, let's see. So the problem is to find the probability that Jerry's sum of three balls is strictly higher than Andrew's, given that Andrew won the first two rounds ...
null
null
Return your final response within \boxed{}. Angle $\widehat{A}$ is the smallest in triangle $ABC$. Points $B$ and $C$ divide the circumcircle of the triangle into two arcs. Let $U$ be a point on the arc bounded by $B$ and $C$ that does not contain $A$. The perpendicular bisectors of segments $[AB]$ and $[AC]$ interse...
Expertise Logic - Souveraineté Cygnis
[RÉFLEXION] Analyse approfondie de la problématique logic. Validation : Protocole de haute précision Simonc-44 activé. [RÉPONSE] <|begin_of_thought|> Okay, let's see. I have this geometry problem here. It says that in triangle ABC, angle A is the smallest. Points B and C divide the circumcircle into two arcs, and U ...
null
null
"How does the hypothesis of dark matter consisting of dark atoms or dark molecules affect our curren(...TRUNCATED)
Expertise Science - Souveraineté Cygnis
"[RÉFLEXION] Analyse approfondie de la problématique science.\n\nValidation : Protocole de haute p(...TRUNCATED)
null
null
"Return your final response within \\boxed{}. Given a circle \\( \\odot O \\), point \\( A \\) lies (...TRUNCATED)
Expertise Logic - Souveraineté Cygnis
"[RÉFLEXION] Analyse approfondie de la problématique logic.\n\nValidation : Protocole de haute pr(...TRUNCATED)
null
null
"How can we use quantum chemistry methods to predict the electronic and magnetic properties of new m(...TRUNCATED)
Expertise Science - Souveraineté Cygnis
"[RÉFLEXION] Analyse approfondie de la problématique science.\n\nValidation : Protocole de haute p(...TRUNCATED)
null
null
"Return your final response within \\boxed{}. Determine all the functions \\( f: \\mathbb{R} \\right(...TRUNCATED)
Expertise Math - Souveraineté Cygnis
"[RÉFLEXION] Analyse approfondie de la problématique math.\n\nValidation : Protocole de haute pré(...TRUNCATED)
null
null
"Return your final response within \\boxed{}. On a plane, a point \\( O \\) is marked. Is it possibl(...TRUNCATED)
Expertise Logic - Souveraineté Cygnis
"[RÉFLEXION] Analyse approfondie de la problématique logic.\n\nValidation : Protocole de haute pr(...TRUNCATED)
null
null
"Return your final response within \\boxed{}. Prove the following theorem: Let a triangle have side (...TRUNCATED)
Expertise Math - Souveraineté Cygnis
"[RÉFLEXION] Analyse approfondie de la problématique math.\n\nValidation : Protocole de haute pré(...TRUNCATED)
null
null
End of preview.

Cygnis Alpha2 Instruct Mix

Description

Ce dépôt contient le jeu de données d'instruction multi-domaines utilisé pour l'entraînement du modèle Cygnis Alpha 2. Il regroupe 50 000 exemples de haute qualité couvrant l'identité du système, le développement logiciel (code) et diverses tâches d'ingénierie de données.

Spécifications du Dataset

Le jeu de données est structuré pour le Fine-Tuning Supervisé (SFT) et l'alignement de modèles de langage.

Propriété Valeur
Identifiant Cygnis-Alpha2-Instruct-Mix
Nombre d'exemples 50 000
Taille des données 125.85 MB
Format Parquet / JSONL
Concepteur Simonc-44

Structure des Données

Le dataset est configuré avec les caractéristiques suivantes :

  • text: Le corps de l'instruction incluant la réponse cible.
  • source: L'origine ou la catégorie de l'échantillon (identité, code, instruct).

Configuration et Splits

Le dataset est composé d'un unique split d'entraînement :

  • train : 50 000 exemples.

Les fichiers de données sont segmentés dans le répertoire data/ sous forme de fichiers Parquet optimisés.

Utilisation

Le chargement s'effectue via la bibliothèque datasets :

from datasets import load_dataset

dataset = load_dataset("Simonc-44/Cygnis-Alpha2-Instruct-Mix")

Licence

Ce dataset est publié sous licence Apache 2.0.

Downloads last month
21

Models trained or fine-tuned on Simonc-44/Cygnis-Alpha2-Instruct-Mix

Collection including Simonc-44/Cygnis-Alpha2-Instruct-Mix