Prediction requirements 
- For questions related to calories, predictions must be given in kilocalories (kcal). 
- For all other nutrient parameters (protein, fat, carbohydrates, etc.) predictions must be given in grams (g). 
- Round predictions up to 2 decimal places.

Constraints 
- You may use any models, but model size must not exceed 4B parameters and the weights must be open. 
- You may not use any additional data beyond what is already provided in the competition. 
- A full run of the solution should be fit within Colab or Kaggle resource limits. 
- You are allowed to use additional tools.

We provide you with MCP servers for <a href="https://mcp-hw-agent.datasouls.com/recipes" rel="noopener noreferrer" target="_blank">Recipe API</a> and <a href="https://mcp-hw-agent.datasouls.com/calories" rel="noopener noreferrer" target="_blank">Calories API</a>.

&nbsp;

You can start from <a href="https://storage.yandexcloud.net/ds-ods/files/files/8082abe6/LLM_agents_start_solution_ipynb_.ipynb" rel="noopener noreferrer" target="_blank">LLM_agents_start_solution_ipynb_.ipynb</a>

sample.csv

test.csv

LLM_agents_start_solution_ipynb_.ipynb

You need to develop a solution to answer questions about the composition of dishes (for example, calories and macronutrients or by recipes), using large language models (LLMs) with support for tool calling. Participants are required to use an open LLM from Hugging Face together with external APIs (a Recipe API and a Calories API) to accurately determine the nutritional value per serving for given dishes.

Public

Agreement

Регистрируясь на курс, я даю <a href="https://storage.yandexcloud.net/ds-ods/files/content/2026/02/09/7ca54807ba64/2025_09_08_%D0%A1%D0%BE%D0%B3%D0%BB%D0%B0%D1%81%D0%B8%D0%B5_%D0%BC%D0%B5%D1%80%D0%BE%D0%BF%D1%80%D0%B8%D1%8F%D1%82%D0%B8%D1%8F_%D0%B8_%D0%BA%D1%83%D1%80%D1%81%D1%8B.pdf">согласие</a> на обработку персональных данных.

first_name

First name

last_name

Last name

email

Email

telegram

Telegram

study_place

Study place

study_specialty

Study specialty

work_place

Work place

work_position

Work position

expectations

Expectations

interests

Interests

preferred_language

Preferred language

about

About

where_from_course

Откуда узнали о курсе?

nlp_student

Если в вашей учебной программе в этом семестре есть курс NLP, выберите свой вуз:

Registration application for Course Season Spring 2026

Practical Assignment 3: Agents

Predictions

Final Projects. NLP-course. Spring 2026

Хакатон

Final Projects

Теория

1. Introduction to Natural Language Processing

Квест

Test 1

Course block deadline expired on 24.02.2026 20:59

Block deadline expired

Materials 1

2. Machine Learning Basics and Text Classification

Test 2

Course block deadline expired on 03.03.2026 20:59

Materials 2

3. Word Embeddings

Test 3

Course block deadline expired on 10.03.2026 20:59

Materials 3

Practice word2vec Lite

Course block deadline expired on 17.03.2026 20:59

Practice Word2Vec Hard

4. Convolutional Neural Networks

Test 4

Materials 4

5. Recurrent Neural Networks and Sequence Tagging

Test 5

Course block deadline expired on 24.03.2026 20:59

Materials 5

6. Machine Translation

Test 6

Course block deadline expired on 31.03.2026 20:59

Materials 6

Practical Assignment 2: Text multiclass classification: store's review rating

Your task is to classify the store's review rating into 5 classes. The metric is F1-score.

Соревнование

Practice 2

Course block deadline expired on 07.04.2026 20:59

7. Transformers

Test 7

Materials 7

8. Large Language Models

Test 8

Course block deadline expired on 14.04.2026 20:59

Materials 8

Practice 3

Course block deadline expired on 28.04.2026 20:59

D. Conversational AI

Test D

Materials D

E. LLM Compression

Materials E

F. Reinforcement Learning for NLP

Materials F

A. Hidden Markov Models and Tagging

Test A

Course block deadline expired on 19.05.2026 20:59

Materials A

B. Topic Modeling

Test B

Materials B

C. Programming Language Processing

Course block will be opened on 25.05.2026 21:00

Block is not open yet

Materials C

Lecturers: 
Valentin Malykh (<a href="http://val.maly.hk/" rel="noopener noreferrer" target="_blank">personal page</a>)

Logistics: 
Lectures at 18.50&nbsp;MSK at Tuesdays online. First lecture will be at 10th of February.&nbsp;

Link to join lectures: <a href="https://my.mts-link.ru/j/MTC/nlp2026spring" rel="noopener noreferrer" target="_blank">https://my.mts-link.ru/j/MTC/nlp2026spring</a>&nbsp;

Seminars will be separate for universities and for ODS. Look info regarding schedule in the chat.

Official support chat: <a href="https://mm.ods.ai/ods/channels/nlp_course" rel="noopener noreferrer" target="_blank">https://mm.ods.ai/ods/channels/nlp_course</a>&nbsp;

Syllabus:

<ul>
	<li>Introduction to Natural Language Processing</li>
	<li>Machine Learning Basics and Text Classification</li>
	<li>Word Embeddings
	<ul>
		<li>1st Assignment</li>
	</ul>
	</li>
	<li>Convolutional Neural Networks
	<ul>
		<li>2nd Assignment</li>
	</ul>
	</li>
	<li>Recurrent Neural Networks</li>
	<li>Statistical Machine Translation</li>
	<li>Transformers</li>
	<li>LLMs</li>
	<li>Project Consultation
	<ul>
		<li>3rd Assignment</li>
	</ul>
	</li>
</ul>

&nbsp;

Prerequisites:

<ul>
	<li>Basic Python Programming</li>
	<li>College Calculus, Linear Algebra</li>
	<li>Basic Probability and Statistics</li>
	<li>Foundations of Machine Learning</li>
</ul>

Assignments:

&nbsp; There will be two assignments on the course.

Projects: 
The participants will be suggested to work on a project during the course. The successful project development is crucial to pass the course. 
The projects from previous runs are available<a href="https://ods.ai/projects" rel="noopener noreferrer" target="_blank"> here</a>.

Natural Language Processing (NLP) is a domain of research whose objective is to analyze and understand human languages and develop technologies to enable human machine interactions with natural languages. NLP is an interdisciplinary field involving linguistics, computer sciences and artificial intelligence. The goal of this course is to provide students with comprehensive knowledge of NLP. Students will be equiped with the principles and theories of NLP, as well as various NLP technologies, including rule-based, statistical and neural network ones. After this course, students will be able to conduct NLP research and develop state-of-the-art NLP systems.

#	Teams (167)	Accuracy Score	Last	Entries
1	p61fe8c278fbb	0.9	6w	1
2	Elena Data Souls	0.78	6w	2
3	Sg Iu	0.78	5w	10
4	Ivan Burmaka	0.76	30d	6
5	Евгений Братковский	0.7	5w	68
6	r820d13c2eeb5	0.7	4w	26
7	Daria Shushkova	0.7	25d	31
8	z19ccecbb1330	0.66	23d	10
9	Микаел Айрапетян	0.64	23d	3
10	Aleksandr Fida	0.62	25d	9
11	m2c362c3e8c5b	0.6	4w	3
12	Радмир Исламов	0.6	23d	12
13	o0746f65adc47	0.56	23d	6
14	Виталий Евтушенко	0.56	26d	2
15	Алексей Красильников	0.56	23d	7
16	x30645c423510	0.56	23d	4
17	Darya Cheburkova	0.54	25d	22
18	Vladimir Bespalov	0.54	28d	72
19	Андрей Богинский	0.54	27d	21
20	Aleksandr Nikitin	0.54	25d	1
21	Kirill Nikulin	0.52	23d	5
22	ndf39247a9a04	0.52	23d	5
23	Anastasia Chernaya	0.5	5w	9
24	Ilya Kustov	0.5	23d	5
25	Daniil Chindin	0.48	30d	9
26	Dima Garanin	0.48	5w	10
27	q1d95e4c690fb	0.48	23d	9
28	Ilya Blinov	0.48	23d	3
29	Nechaeva Kseniia	0.48	23d	5
30	bd9d27e0377d4	0.48	23d	5
31	z59252e5a218f	0.48	23d	1
32	Vsevolod Ershov	0.46	5w	7
33	Maxim Mikhalevich	0.44	5w	7
34	l9254cb101c2c	0.44	29d	1
35	Александр Губич	0.44	24d	10
36	Константин Румовский	0.42	23d	24
37	je76a57537fa3	0.42	26d	7
38	fggnhguocdo	0.42	24d	1
39	Lidiya S	0.38	29d	21
40	c52b40dba634d	0.38	23d	13
41	pb045208f92b0	0.38	23d	5
42	Роман Вяткин	0.36	29d	63
43	Elizaveta Petrusevich	0.36	25d	34
44	dc9e614a76a3c	0.36	26d	30
45	y3f93f3e8c6bf	0.34	5w	62
46	Ivan Papai	0.34	5w	24
47	Yuri Fokichev	0.34	23d	28
48	Nick Linker	0.34	4w	47
49	Danil Kornovan	0.34	24d	35
50	z9ff848d2fd0f	0.34	24d	6