Llama-2-13b-chat-hf

MERA Created at 22.09.2024 20:42
0.249
The overall result
538
Place in the rating
Weak tasks:
301
RWSD
515
PARus
616
RCB
503
ruEthics
470
MultiQ
518
ruWorldTree
530
ruOpenBookQA
326
CheGeKa
552
ruMMLU
512
ruHateSpeech
398
ruDetox
393
ruHHH
472
ruTiE
607
USE
519
MathLogicQA
481
ruMultiAr
411
SimpleAr
502
LCS
443
BPS
530
MaMuRAMu
+16
Hide

Ratings for leaderboard tasks

The table will scroll to the left

Task name Result Metric
LCS 0.062 Accuracy
RCB 0.308 / 0.246 Accuracy F1 macro
USE 0.009 Grade norm
RWSD 0.527 Accuracy
PARus 0.51 Accuracy
ruTiE 0.512 Accuracy
MultiQ 0.223 / 0.036 F1 Exact match
CheGeKa 0.071 / 0.048 F1 Exact match
ruModAr 0 Exact match
MaMuRAMu 0.366 Accuracy
ruMultiAr 0.096 Exact match
ruCodeEval 0 / 0 / 0 Pass@k
MathLogicQA 0.283 Accuracy
ruWorldTree 0.51 / 0.491 Accuracy F1 macro
ruOpenBookQA 0.415 / 0.383 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 0.692 Accuracy
ruMMLU 0.303 Accuracy
SimpleAr 0.833 Exact match
ruHumanEval 0 / 0 / 0 Pass@k
ruHHH 0.551
ruHateSpeech 0.513
ruDetox 0.116
ruEthics
Correct God Ethical
Virtue 0.015 0.091 0.105
Law 0.051 0.121 0.116
Moral 0.05 0.112 0.133
Justice 0.021 0.094 0.103
Utilitarianism 0.036 0.091 0.122

Information about the submission

Mera version
v.1.2.0
Torch Version
2.4.0
The version of the codebase
9b26db97
CUDA version
12.1
Precision of the model weights
float16
Seed
1234
Batch
1
Transformers version
4.44.2
The number of GPUs and their type
2 x NVIDIA H100 80GB HBM3
Architecture
vllm

Team:

MERA

Name of the ML model:

Llama-2-13b-chat-hf

Model size

13.0B

Model type:

Opened

SFT

Additional links:

https://arxiv.org/pdf/2307.09288

Architecture description:

Auto-regressive language model that uses an optimized transformer architecture, with context length of 4K tokens.

Description of the training:

Pretrained and fine-tuned (SFT & RLHF) generative text model, optimized for dialogue use cases. The global batch-size is 4M tokens. The model was trained between January 2023 and July 2023.

Pretrain data:

A new mix of publicly available online data, the pretraining data volume is 2.0T.

License:

https://ai.meta.com/resources/models-and-libraries/llama-downloads/

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, bps, lcs, chegeka, mathlogicqa, parus, rcb, rudetox, ruhatespeech, rummlu, ruworldtree, ruopenbookqa, rumultiar, use, rwsd, mamuramu, multiq, rumodar, ruethics, ruhhh, ruhumaneval, rucodeeval, rutie, rutie - 4096

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-2-13b-chat-hf
MERA
0 0.067 0 0.1 0 0.067 0 - 0 0 0 0 0.067 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Model, team Honest Helpful Harmless
Llama-2-13b-chat-hf
MERA
0.574 0.525 0.552
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-2-13b-chat-hf
MERA
0.333 0.277 0.263 0.415 0.281 0.333 0.379 0.36 0.34 0.327 0.193 0.238 0.36 0.37 0.39 0.361 0.35 0.264 0.2 0.313 0.28 0.415 0.2 0.324 0.196 0.3 0.258 0.259 0.196 0.32 0.28 0.455 0.282 0.354 0.298 0.269 0.33 0.371 0.258 0.315 0.354 0.342 0.248 0.257 0.326 0.301 0.289 0.244 0.305 0.277 0.31 0.262 0.31 0.261 0.31 0.242 0.363
Model, team SIM FL STA
Llama-2-13b-chat-hf
MERA
0.455 0.631 0.54
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-2-13b-chat-hf
MERA
0.378 0.327 0.333 0.343 0.329 0.276 0.397 0.333 0.288 0.4 0.308 0.275 0.283 0.403 0.263 0.481 0.28 0.356 0.386 0.509 0.211 0.407 0.267 0.385 0.356 0.394 0.372 0.246 0.474 0.378 0.756 0.372 0.375 0.368 0.364 0.464 0.311 0.511 0.228 0.6 0.42 0.365 0.378 0.244 0.31 0.489 0.448 0.227 0.462 0.316 0.289 0.333 0.43 0.286 0.349 0.322 0.567
Coorect
Good
Ethical
Model, team Virtue Law Moral Justice Utilitarianism
Llama-2-13b-chat-hf
MERA
0.015 0.051 0.05 0.021 0.036
Model, team Virtue Law Moral Justice Utilitarianism
Llama-2-13b-chat-hf
MERA
0.091 0.121 0.112 0.094 0.091
Model, team Virtue Law Moral Justice Utilitarianism
Llama-2-13b-chat-hf
MERA
0.105 0.116 0.133 0.103 0.122
Model, team Women Men LGBT Nationalities Migrants Other
Llama-2-13b-chat-hf
MERA
0.537 0.457 0.353 0.514 0.429 0.557