RU EN
Return to the leaderboard

ruadapt llama3-8B-instruct lep ft

RCC MSU Created at 26.09.2024 16:53
0.447
The overall result
195
Place in the rating

Ratings for leaderboard tasks

The table will scroll to the left

Task name Result Metric
LCS 0.18 Accuracy
RCB 0.534 / 0.446 Accuracy F1 macro
USE 0.124 Grade norm
RWSD 0.542 Accuracy
PARus 0.828 Accuracy
ruTiE 0.707 Accuracy
MultiQ 0.483 / 0.334 F1 Exact match
CheGeKa 0.146 / 0.101 F1 Exact match
ruModAr 0.454 Exact match
MaMuRAMu 0.633 Accuracy
ruMultiAr 0.244 Exact match
ruCodeEval 0 / 0 / 0 Pass@k
MathLogicQA 0.362 Accuracy
ruWorldTree 0.838 / 0.837 Accuracy F1 macro
ruOpenBookQA 0.773 / 0.772 Accuracy F1 macro

Evaluation on open tasks:

Go to the ratings by subcategory

The table will scroll to the left

Task name Result Metric
BPS 0.947 Accuracy
ruMMLU 0.52 Accuracy
SimpleAr 0.941 Exact match
ruHumanEval 0.009 / 0.012 / 0.012 Pass@k
ruHHH 0.713
ruHateSpeech 0.706
ruDetox 0.257
ruEthics
Correct God Ethical
Virtue 0.363 0.488 0.454
Law 0.421 0.48 0.468
Moral 0.41 0.496 0.501
Justice 0.33 0.425 0.425
Utilitarianism 0.332 0.432 0.393

Information about the submission:

Mera version
v.1.2.0
Torch Version
2.3.0
The version of the codebase
430295f
CUDA version
12.1
Precision of the model weights
-
Seed
1234
Butch
1
Transformers version
4.44.2
The number of GPUs and their type
NVIDIA A100
Architecture
vllm

Team:

RCC MSU

Name of the ML model:

ruadapt llama3-8B-instruct lep ft

Link to the ML model:

https://huggingface.co/RefalMachine/ruadapt_llama3_8b_instruct_extended_lep_ft

Model size

8.4B

Model type:

Opened

SFT

Architecture description:

LoRa tuned version of ruadapt llama 3 8B with extended tokenizer after LEP (Learned Embedding Propagation, paper will be soon) procedure on saiga_scored d7 dataset. Thanks to the extended tokenizer, the model works more efficiently with the Russian language. How to cite: Tikhomirov M., Chernyshev D. Facilitating large language model Russian adaptation with Learned Embedding Propagation // 2024 (will be soon) Tikhomirov M., Chernyshev D. Impact of Tokenization on LLaMa Russian Adaptation //2023 Ivannikov Ispras Open Conference (ISPRAS). – IEEE, 2023. – С. 163-168.

Inference parameters

Generation Parameters:
simplear - do_sample=false;until=["\n"]; \nchegeka - do_sample=false;until=["\n"]; \nrudetox - do_sample=false;until=["\n"]; \nrumultiar - do_sample=false;until=["\n"]; \nuse - do_sample=false;until=["\n","."]; \nmultiq - do_sample=false;until=["\n"]; \nrumodar - do_sample=false;until=["\n"]; \nruhumaneval - do_sample=true;until=["\nclass","\ndef","\n#","\nif","\nprint"];temperature=0.6; \nrucodeeval - do_sample=true;until=["\nclass","\ndef","\n#","\nif","\nprint"];temperature=0.6;

The size of the context:
simplear, chegeka, rudetox, rumultiar, use, multiq, rumodar, ruhumaneval, rucodeeval - 8192

System prompt:
Реши задачу по инструкции ниже. Не давай никаких объяснений и пояснений к своему ответу. Не пиши ничего лишнего. Пиши только то, что указано в инструкции. Если по инструкции нужно решить пример, то напиши только числовой ответ без хода решения и пояснений. Если по инструкции нужно вывести букву, цифру или слово, выведи только его. Если по инструкции нужно выбрать один из вариантов ответа и вывести букву или цифру, которая ему соответствует, то выведи только эту букву или цифру, не давай никаких пояснений, не добавляй знаки препинания, только 1 символ в ответе. Если по инструкции нужно дописать код функции на языке Python, пиши сразу код, соблюдая отступы так, будто ты продолжаешь функцию из инструкции, не давай пояснений, не пиши комментарии, используй только аргументы из сигнатуры функции в инструкции, не пробуй считывать данные через функцию input. Не извиняйся, не строй диалог. Выдавай только ответ и ничего больше.

Description of the template:
{% set loop_messages = messages %}{% for message in loop_messages %}{% set content = '<|start_header_id|>' + message['role'] + '<|end_header_id|> \n \n'+ message['content'] | trim + '<|eot_id|>' %}{% if loop.index0 == 0 %}{% set content = bos_token + content %}{% endif %}{{ content }}{% endfor %}{% if add_generation_prompt %}{{ '<|start_header_id|>assistant<|end_header_id|> \n \n' }}{% endif %}

Expand information

Ratings by subcategory

Metric: Grade Norm
Model, team 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 8_0 8_1 8_2 8_3 8_4
ruadapt llama3-8B-instruct lep ft
RCC MSU 		0.167 0.133 0.733 0.267 0 0.1 0 - 0 0 0.067 0.067 0.1 0 0.133 0.267 0 0.033 0 0.033 0 0.2 0.067 0.033 0.067 0.117 0.033 0.067 0.467 0.167 0.267
Model, team Honest Helpful Harmless
ruadapt llama3-8B-instruct lep ft
RCC MSU 		0.738 0.661 0.741
Model, team Anatomy Virology Astronomy Marketing Nutrition Sociology Management Philosophy Prehistory Human aging Econometrics Formal logic Global facts Jurisprudence Miscellaneous Moral disputes Business ethics Biology (college) Physics (college) Human Sexuality Moral scenarios World religions Abstract algebra Medicine (college) Machine learning Medical genetics Professional law PR Security studies Chemistry (школьная) Computer security International law Logical fallacies Politics Clinical knowledge Conceptual_physics Math (college) Biology (high school) Physics (high school) Chemistry (high school) Geography (high school) Professional medicine Electrical engineering Elementary mathematics Psychology (high school) Statistics (high school) History (high school) Math (high school) Professional accounting Professional psychology Computer science (college) World history (high school) Macroeconomics Microeconomics Computer science (high school) European history Government and politics
ruadapt llama3-8B-instruct lep ft
RCC MSU 		0.504 0.428 0.651 0.714 0.605 0.721 0.65 0.588 0.556 0.587 0.351 0.389 0.37 0.602 0.637 0.578 0.68 0.514 0.333 0.649 0.256 0.684 0.31 0.457 0.393 0.56 0.375 0.574 0.686 0.36 0.62 0.76 0.479 0.737 0.581 0.5 0.35 0.668 0.43 0.473 0.677 0.445 0.634 0.414 0.661 0.394 0.662 0.359 0.401 0.511 0.46 0.722 0.51 0.576 0.69 0.703 0.684
Model, team SIM FL STA
ruadapt llama3-8B-instruct lep ft
RCC MSU 		0.692 0.65 0.634
Model, team Anatomy Virology Astronomy Marketing Nutrition Sociology Managment Philosophy Pre-History Gerontology Econometrics Formal logic Global facts Jurisprudence Miscellaneous Moral disputes Business ethics Bilology (college) Physics (college) Human sexuality Moral scenarios World religions Abstract algebra Medicine (college) Machine Learning Genetics Professional law PR Security Chemistry (college) Computer security International law Logical fallacies Politics Clinical knowledge Conceptual physics Math (college) Biology (high school) Physics (high school) Chemistry (high school) Geography (high school) Professional medicine Electrical Engineering Elementary mathematics Psychology (high school) Statistics (high school) History (high school) Math (high school) Professional Accounting Professional psychology Computer science (college) World history (high school) Macroeconomics Microeconomics Computer science (high school) Europe History Government and politics
ruadapt llama3-8B-instruct lep ft
RCC MSU 		0.467 0.624 0.6 0.537 0.645 0.69 0.5 0.614 0.692 0.554 0.654 0.625 0.492 0.659 0.608 0.654 0.664 0.622 0.421 0.702 0.298 0.695 0.556 0.675 0.622 0.667 0.641 0.579 0.842 0.667 0.778 0.654 0.607 0.789 0.5 0.536 0.533 0.733 0.421 0.677 0.686 0.571 0.822 0.622 0.897 0.8 0.793 0.614 0.615 0.807 0.778 0.667 0.684 0.558 0.465 0.567 0.744
Coorect
Good
Ethical
Model, team Virtue Law Moral Justice Utilitarianism
ruadapt llama3-8B-instruct lep ft
RCC MSU 		0.363 0.421 0.41 0.33 0.332
Model, team Virtue Law Moral Justice Utilitarianism
ruadapt llama3-8B-instruct lep ft
RCC MSU 		0.488 0.48 0.496 0.425 0.432
Model, team Virtue Law Moral Justice Utilitarianism
ruadapt llama3-8B-instruct lep ft
RCC MSU 		0.454 0.468 0.501 0.425 0.393
Model, team Women Men LGBT Nationalities Migrants Other
ruadapt llama3-8B-instruct lep ft
RCC MSU 		0.75 0.686 0.706 0.649 0.571 0.689

Пользовательское соглашение

We use cookies to store data. By continuing to use the site, you consent to work with these files