T-lite-0.1

T-Bank AI Created at 16.08.2024 09:45

0.492

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	0.14	Accuracy
RCB	0.511 / 0.418	Accuracy F1 macro
USE	0.05	Grade norm
RWSD	0.585	Accuracy
PARus	0.858	Accuracy
ruTiE	0.681	Accuracy
MultiQ	0.383 / 0.29	F1 Exact match
CheGeKa	0.118 / 0.06	F1 Exact match
ruModAr	0.667	Exact match
ruMultiAr	0.269	Exact match
MathLogicQA	0.37	Accuracy
ruWorldTree	0.88 / 0.88	Accuracy F1 macro
ruOpenBookQA	0.783 / 0.782	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

0.358

Accuracy

ruMMLU

0.759

Accuracy

SimpleAr

0.955

Exact match

ruHumanEval

0.023 / 0.113 / 0.226

Pass@k

ruHHH

0.596

ruHateSpeech

0.732

ruDetox

0.197

ruEthics

	Correct	God	Ethical
Virtue	-0.076	-0.187	-0.225
Law	-0.091	-0.213	-0.23
Moral	-0.092	-0.193	-0.222
Justice	-0.071	-0.185	-0.215
Utilitarianism	-0.11	-0.153	-0.231

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:

T-Bank AI

Name of the ML model:

T-lite-0.1

Link to the ML model:

https://huggingface.co/AnatoliiPotapov/T-lite-0.1

Additional links:

https://huggingface.co/AnatoliiPotapov/T-lite-instruct-0.1

Architecture description:

T-lite is a decoder language model with: pre-normalization via RMSNorm SwiGLU activation function rotary positional embeddings (RoPE) grouped query attention (GQA) T-lite was trained in bf16.

Description of the training:

We employed the Decoupled AdamW optimizer with β1 = 0.9, β2 = 0.95, and eps = 1.0e-8. The learning rate was set to 1.0e-5 with a constant schedule and a warmup period of 10 steps during stage 1, and a cosine schedule during stage 2. Weight decay was applied at a rate of 1.0e-6, and gradient clipping was performed with a maximum norm of 1.0. The maximum sequence length was set to 8192. Each batch contained approximately 6 million tokens. Training was conducted using Fully Sharded Data Parallel (FSDP) with full shard/hybrid shard strategies. The setup achieved a throughput of 3000 tokens/sec/GPU. We achieved a 0.59 Model FLOPs Utilization (MFU).

Pretrain data:

Stage 1 Massive continual pre-training 300B tokens * 0.3 epoch Proportion of data in Russian is 85%, as a trade-off between language adoptation and English language performance Styles and topics in Common Crawl (CC) data were downsampled Domains in book datasets were balanced Proportion of code data was increased Stage 2 Focuses on refining the quality of the dataset 20B tokens * 3 epochs Includes instructional sets of smaller volume Advertisements and news were aggressively downsampled Instructions and articles were upsampled Educational content was balanced

Training Details:

License:

WTFPL

Strategy, generation and parameters:

1 Nvidia A100 Context length: 8192 dtype: bfloat16 Pytorch==2.3.1 + Transformers 4.44.0 + CUDA 12.1

Comments about inference:

🚨 T-lite is designed for further fine-tuning and is not intended as a ready-to-use conversational assistant. Users are advised to exercise caution and are responsible for any additional training and oversight required to ensure the model's responses meet acceptable ethical and safety standards. The responsibility for incorporating this model into industrial or commercial solutions lies entirely with those who choose to deploy it.

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
T-lite-0.1 T-Bank AI	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-

Model, team	Honest	Helpful	Harmless
T-lite-0.1 T-Bank AI	0.541	0.542	0.707

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
T-lite-0.1 T-Bank AI	0.8	0.813	0.9	0.771	0.762	1	0.8	0.824	0.8	1	0.818	0.9	0.7	0.654	0.5	0.5	0.7	0.815	0.7	1	0.4	0.808	0.8	0.765	0.7	0.909	0.688	0.714	0.9	0.818	0.6	0.889	0.6	0.8	0.818	1	1	0.81	0.5	0.6	0.835	0.9	0.9	0.9	0.938	0.7	0.9	0.5	0.5	1	0.5	0.813	0.912	0.867	0.333	0.515	0.704

Model, team	SIM	FL	STA
T-lite-0.1 T-Bank AI	0.511	0.727	0.419

Coorect

Good

Ethical

Model, team	Virtue	Law	Moral	Justice	Utilitarianism
T-lite-0.1 T-Bank AI	-0.076	-0.091	-0.092	-0.071	-0.11

Model, team	Virtue	Law	Moral	Justice	Utilitarianism
T-lite-0.1 T-Bank AI	-0.187	-0.213	-0.193	-0.185	-0.153

Model, team	Virtue	Law	Moral	Justice	Utilitarianism
T-lite-0.1 T-Bank AI	-0.225	-0.23	-0.222	-0.215	-0.231

Model, team	Women	Men	LGBT	Nationalities	Migrants	Other
T-lite-0.1 T-Bank AI	0.731	0.743	0.706	0.649	0.571	0.803

T-lite-0.1

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:

Comments about inference:

Ratings by subcategory

Confirm the deletion of the sub