ruGPT-3.5 13B

MERA Created at 12.01.2024 11:18

0.208

The overall result

The submission does not contain all the required tasks

Ratings for leaderboard tasks

The table will scroll to the left

Task name	Result	Metric
LCS	0.132	Accuracy
RCB	0.331 / 0.194	Accuracy F1 macro
USE	0.025	Grade norm
RWSD	0.523	Accuracy
PARus	0.504	Accuracy
ruTiE	0.488	Accuracy
MultiQ	0.115 / 0.036	F1 Exact match
CheGeKa	0.037 / 0	F1 Exact match
ruModAr	0.001	Exact match
ruMultiAr	0.025	Exact match
MathLogicQA	0.258	Accuracy
ruWorldTree	0.246 / 0.22	Accuracy F1 macro
ruOpenBookQA	0.223 / 0.208	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

0.492

Accuracy

ruMMLU

0.246

Accuracy

SimpleAr

0.029

Exact match

ruHumanEval

0.001 / 0.003 / 0.006

Pass@k

ruHHH

0.472

ruHateSpeech

0.543

ruDetox

0.286

ruEthics

	Correct	God	Ethical
Virtue	-0.036	0.045	0.034
Law	-0.023	0.035	-0.021
Moral	-0.025	0.034	0.029
Justice	-0.017	0.045	0.049
Utilitarianism	-0.016	0.04	0.067

Information about the submission

Mera version

Torch Version

The version of the codebase

CUDA version

Precision of the model weights

Seed

Batch

Transformers version

The number of GPUs and their type

Architecture

Team:

MERA

Name of the ML model:

ruGPT-3.5 13B

Link to the ML model:

https://huggingface.co/ai-forever/ruGPT-3.5-13B

Additional links:

https://habr.com/ru/companies/sberbank/articles/746736/

Architecture description:

ruGPT-3 is a Russian counterpart of GPT-3 (Brown et al., 2020). Model has 13B parameters. This is the biggest model so far and it was used for training first version of GigaChat.

Description of the training:

Model was trained using Deepspeed and Megatron libraries, on 300B tokens dataset for 3 epochs, around 45 days on 512 V100. After that model was finetuned 1 epoch with sequence length 2048 around 20 days on 200 GPU A100 on additional data (see above).

Pretrain data:

Model was pretrained on a 300Gb of various domains, than additionaly trained on the 100 Gb of code and legal documents. Training data was deduplicated, the text deduplication includes 64-bit hashing of each text in the corpus for keeping texts with a unique hash. We also filter the documents based on their text compression rate using zlib4. The most strongly and weakly compressing deduplicated texts are discarded.

Training Details:

After the final training perplexity for this model was around 8.8 for Russian.

License:

MIT

Strategy, generation and parameters:

Code version v.1.1.0 All the parameters were not changed and are used as prepared by the organizers. Details: - 1 x NVIDIA A100 - dtype auto - Pytorch 2.1.2 + CUDA 12.1 - Transformers 4.36.2 - Context length 2048

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
ruGPT-3.5 13B MERA	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-

Model, team	Honest	Helpful	Harmless
ruGPT-3.5 13B MERA	0.475	0.475	0.466

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
ruGPT-3.5 13B MERA	0.2	0.25	0.3	0.343	0.238	0	0.267	0.294	0.2	0.4	0.455	0.3	0.1	0.269	0.136	0.1	0.1	0.259	0.4	0.2	0.2	0.269	0.1	0.333	0.2	0.455	0.313	0.214	0.1	0.182	0.2	0.333	0.1	0.3	0.182	0.2	0.4	0.381	0.3	0.2	0.203	0.4	0.3	0.2	0.25	0.2	0.4	0	0.1	0.1	0.318	0.313	0.176	0.333	0.208	0.212	0.148

Model, team	SIM	FL	STA
ruGPT-3.5 13B MERA	0.562	0.704	0.678

Coorect

Good

Ethical

Model, team	Virtue	Law	Moral	Justice	Utilitarianism
ruGPT-3.5 13B MERA	-0.036	-0.023	-0.025	-0.017	-0.016

Model, team	Virtue	Law	Moral	Justice	Utilitarianism
ruGPT-3.5 13B MERA	0.045	0.035	0.034	0.045	0.04

Model, team	Virtue	Law	Moral	Justice	Utilitarianism
ruGPT-3.5 13B MERA	0.034	-0.021	0.029	0.049	0.067

Model, team	Women	Men	LGBT	Nationalities	Migrants	Other
ruGPT-3.5 13B MERA	0.537	0.657	0.647	0.514	0.286	0.508

ruGPT-3.5 13B

Ratings for leaderboard tasks

Evaluation on open tasks:

Information about the submission

Team:

Name of the ML model:

Link to the ML model:

Additional links:

Architecture description:

Description of the training:

Pretrain data:

Training Details:

License:

Strategy, generation and parameters:

Ratings by subcategory

Confirm the deletion of the sub