gemma-2-2b-it

MERA Created at 22.09.2024 22:14
0.391
The overall result
347
Place in the rating
Weak tasks:
271
RWSD
461
PARus
427
RCB
424
ruEthics
485
MultiQ
360
ruWorldTree
318
ruOpenBookQA
377
CheGeKa
406
ruMMLU
420
ruHateSpeech
290
ruDetox
381
ruHHH
377
ruTiE
76
ruHumanEval
493
USE
369
MathLogicQA
338
ruMultiAr
337
SimpleAr
393
LCS
486
BPS
306
ruModAr
400
MaMuRAMu
75
ruCodeEval
+19
Hide

Ratings for leaderboard tasks

The table will scroll to the left

Task name Result Metric
LCS 0.088 Accuracy
RCB 0.463 / 0.437 Accuracy F1 macro
USE 0.058 Grade norm
RWSD 0.535 Accuracy
PARus 0.556 Accuracy
ruTiE 0.567 Accuracy
MultiQ 0.195 / 0.143 F1 Exact match
CheGeKa 0.041 / 0.024 F1 Exact match
ruModAr 0.455 Exact match
MaMuRAMu 0.507 Accuracy
ruMultiAr 0.221 Exact match
ruCodeEval 0.29 / 0.471 / 0.524 Pass@k
MathLogicQA 0.357 Accuracy
ruWorldTree 0.75 / 0.75 Accuracy F1 macro
ruOpenBookQA 0.693 / 0.693 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 0.583 Accuracy
ruMMLU 0.43 Accuracy
SimpleAr 0.918 Exact match
ruHumanEval 0.271 / 0.399 / 0.439 Pass@k
ruHHH 0.562
ruHateSpeech 0.558
ruDetox 0.164
ruEthics
Correct God Ethical
Virtue 0.099 0.181 0.15
Law 0.088 0.155 0.099
Moral 0.121 0.174 0.145
Justice 0.101 0.155 0.13
Utilitarianism 0.085 0.109 0.141

Information about the submission

Mera version
v.1.2.0
Torch Version
2.3.1
The version of the codebase
aec92f8
CUDA version
12.1
Precision of the model weights
bfloat16
Seed
1234
Batch
1
Transformers version
4.43.3
The number of GPUs and their type
1 x NVIDIA H100 80GB HBM3
Architecture
hf

Team:

MERA

Name of the ML model:

gemma-2-2b-it

Model size

2.6B

Model type:

Opened

SFT

Architecture description:

Gemma is built from the same research and technology used to create the Gemini models. They are text-to-text, decoder-only large language models, available in English, with open weights for both pre-trained variants and instruction-tuned variants

Description of the training:

-

Pretrain data:

These models were trained on a dataset of text data that includes a wide variety of sources. The models d was trained with 2 trillion tokens.

License:

gemma

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, rutie, ruhumaneval, rucodeeval - 8192

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
gemma-2-2b-it
MERA
0.067 0.1 0.267 0.1 0 0.2 0 - 0 0 0 0 0.033 0 0 0.15 0.033 0 0 0 0.033 0.233 0 0 0 0.042 0.067 0.067 0.167 0.033 0.1
Model, team Honest Helpful Harmless
gemma-2-2b-it
MERA
0.492 0.542 0.655
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
gemma-2-2b-it
MERA
0.422 0.416 0.428 0.65 0.458 0.572 0.592 0.457 0.475 0.493 0.254 0.389 0.29 0.519 0.489 0.477 0.46 0.375 0.233 0.534 0.245 0.392 0.31 0.416 0.232 0.48 0.346 0.509 0.592 0.39 0.58 0.678 0.429 0.657 0.472 0.372 0.39 0.529 0.358 0.369 0.535 0.342 0.503 0.379 0.537 0.38 0.534 0.293 0.355 0.416 0.37 0.591 0.444 0.441 0.53 0.558 0.451
Model, team SIM FL STA
gemma-2-2b-it
MERA
0.375 0.727 0.705
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
gemma-2-2b-it
MERA
0.311 0.475 0.367 0.519 0.434 0.552 0.466 0.421 0.404 0.554 0.538 0.442 0.4 0.566 0.357 0.346 0.607 0.489 0.351 0.491 0.316 0.576 0.422 0.574 0.489 0.485 0.577 0.456 0.825 0.578 0.711 0.513 0.527 0.526 0.455 0.518 0.4 0.756 0.333 0.538 0.559 0.54 0.489 0.511 0.862 0.756 0.397 0.636 0.631 0.737 0.556 0.464 0.582 0.481 0.419 0.351 0.644
Coorect
Good
Ethical
Model, team Virtue Law Moral Justice Utilitarianism
gemma-2-2b-it
MERA
0.099 0.088 0.121 0.101 0.085
Model, team Virtue Law Moral Justice Utilitarianism
gemma-2-2b-it
MERA
0.181 0.155 0.174 0.155 0.109
Model, team Virtue Law Moral Justice Utilitarianism
gemma-2-2b-it
MERA
0.15 0.099 0.145 0.13 0.141
Model, team Women Men LGBT Nationalities Migrants Other
gemma-2-2b-it
MERA
0.574 0.486 0.529 0.622 0.286 0.574