Llama-3.3-70B-Instruct

llmarena.ru Created at 17.03.2025 08:12
0.555
The overall result
104
Place in the rating
Weak tasks:
99
RWSD
79
PARus
77
RCB
121
ruEthics
69
MultiQ
133
ruWorldTree
88
ruOpenBookQA
94
CheGeKa
76
ruMMLU
171
ruHateSpeech
65
ruDetox
106
ruHHH
89
ruTiE
149
ruHumanEval
103
USE
103
MathLogicQA
117
ruMultiAr
249
SimpleAr
70
LCS
160
BPS
150
ruModAr
97
MaMuRAMu
131
ruCodeEval
+19
Hide

Ratings for leaderboard tasks

The table will scroll to the left

Task name Result Metric
LCS 0.168 Accuracy
RCB 0.575 / 0.38 Accuracy F1 macro
USE 0.298 Grade norm
RWSD 0.6 Accuracy
PARus 0.914 Accuracy
ruTiE 0.824 Accuracy
MultiQ 0.573 / 0.418 F1 Exact match
CheGeKa 0.339 / 0.276 F1 Exact match
ruModAr 0.57 Exact match
MaMuRAMu 0.802 Accuracy
ruMultiAr 0.34 Exact match
ruCodeEval 0.139 / 0.28 / 0.396 Pass@k
MathLogicQA 0.566 Accuracy
ruWorldTree 0.954 / 0.769 Accuracy F1 macro
ruOpenBookQA 0.91 / 0.735 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 0.965 Accuracy
ruMMLU 0.748 Accuracy
SimpleAr 0.959 Exact match
ruHumanEval 0.116 / 0.27 / 0.378 Pass@k
ruHHH 0.826
ruHateSpeech 0.774
ruDetox 0.322
ruEthics
Correct God Ethical
Virtue 0.373 0.36 0.417
Law 0.362 0.351 0.403
Moral 0.41 0.386 0.447
Justice 0.348 0.316 0.375
Utilitarianism 0.315 0.324 0.379

Information about the submission

Mera version
v.1.2.0
Torch Version
2.5.1
The version of the codebase
30667dc
CUDA version
12.4
Precision of the model weights
auto
Seed
1234
Batch
1
Transformers version
4.49.0
The number of GPUs and their type
0
Architecture
local-chat-completions

Team:

llmarena.ru

Name of the ML model:

Llama-3.3-70B-Instruct

Model size

70.0B

Model type:

Closed

API

SFT

Additional links:

Замер осуществлялся через сервис openrouter (https://openrouter.ai/meta-llama/llama-3.3-70b-instruct)

Architecture description:

The Meta Llama 3.3 multilingual large language model (LLM) is an instruction tuned generative model in 70B (text in/text out). The Llama 3.3 instruction tuned text only model is optimized for multilingual dialogue use cases and outperforms many of the available open source and closed chat models on common industry benchmarks.

Description of the training:

Llama 3.3 is an auto-regressive language model that uses an optimized transformer architecture. The tuned versions use supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF) to align with human preferences for helpfulness and safety.

Pretrain data:

Llama 3.3 was pretrained on ~15 trillion tokens of data from publicly available sources. The fine-tuning data includes publicly available instruction datasets, as well as over 25M synthetically generated examples.

License:

Llama 3.3 Community License Agreement (https://github.com/meta-llama/llama-models/blob/main/models/llama3_3/LICENSE)

Inference parameters

Generation Parameters:
rucodeeval - do_sample=true;temperature=0.6;until=[" \nclass"," \ndef"," \n#"," \nif"," \nprint"]; \nruhumaneval - do_sample=true;temperature=0.6;until=[" \nclass"," \ndef"," \n#"," \nif"," \nprint"];

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.

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
Llama-3.3-70B-Instruct
llmarena.ru
0.467 0.433 0.833 0.2 0.1 0.433 0.1 - 0.1 0.033 0.033 0.067 0.3 0.1 0 0.483 0.033 0 0 0.033 0.033 0.633 0.4 0.367 0.1 0.608 0.2 0.4 0.433 0.4 0.5
Model, team Honest Helpful Harmless
Llama-3.3-70B-Instruct
llmarena.ru
0.82 0.78 0.879
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
Llama-3.3-70B-Instruct
llmarena.ru
0.689 0.56 0.836 0.859 0.81 0.846 0.845 0.765 0.827 0.789 0.588 0.579 0.61 0.824 0.829 0.775 0.75 0.875 0.511 0.771 0.716 0.848 0.53 0.728 0.634 0.89 0.593 0.694 0.8 0.53 0.8 0.884 0.748 0.919 0.762 0.748 0.53 0.89 0.603 0.69 0.864 0.868 0.717 0.708 0.88 0.662 0.907 0.507 0.553 0.752 0.7 0.882 0.815 0.866 0.87 0.861 0.912
Model, team SIM FL STA
Llama-3.3-70B-Instruct
llmarena.ru
0.693 0.655 0.745
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
Llama-3.3-70B-Instruct
llmarena.ru
0.733 0.881 0.717 0.713 0.868 0.828 0.69 0.737 0.942 0.723 0.795 0.75 0.6 0.775 0.842 0.728 0.673 0.778 0.719 0.842 0.842 0.881 0.844 0.852 0.867 0.848 0.795 0.754 0.895 0.756 0.822 0.846 0.75 0.912 0.697 0.875 0.822 0.844 0.649 0.769 0.882 0.905 0.822 0.911 0.897 0.911 0.879 0.841 0.8 0.947 0.867 0.913 0.759 0.753 0.605 0.743 0.867
Coorect
Good
Ethical
Model, team Virtue Law Moral Justice Utilitarianism
Llama-3.3-70B-Instruct
llmarena.ru
0.373 0.362 0.41 0.348 0.315
Model, team Virtue Law Moral Justice Utilitarianism
Llama-3.3-70B-Instruct
llmarena.ru
0.36 0.351 0.386 0.316 0.324
Model, team Virtue Law Moral Justice Utilitarianism
Llama-3.3-70B-Instruct
llmarena.ru
0.417 0.403 0.447 0.375 0.379
Model, team Women Men LGBT Nationalities Migrants Other
Llama-3.3-70B-Instruct
llmarena.ru
0.796 0.686 0.588 0.838 0.714 0.803