mGPT 13B
Создан 12.01.2024 11:17Общая оценка: 0.196
Таблица скроллится влево
| Задача | Результат | Метрика |
|---|---|---|
| BPS | 0.463 | Accuracy |
| LCS | 0.132 | Accuracy |
| RCB | 0.333 / 0.167 | Avg. F1 / Accuracy |
| USE | 0.002 | Grade Norm |
| RWSD | 0.485 | Accuracy |
| PARus | 0.498 | Accuracy |
| ruTiE | 0.5 | Accuracy |
| MultiQ | 0.062 / 0.023 | F1-score/EM |
| ruMMLU | 0.235 | Accuracy |
| CheGeKa | 0.006 / 0 | F1 / EM |
| ruModAr | 0.0 | Accuracy |
| SimpleAr | 0.023 | Accuracy |
| ruMultiAr | 0.019 | Accuracy |
| MathLogicQA | 0.263 | Accuracy |
| ruHumanEval | 0 / 0 / 0 | pass@k |
| ruWorldTree | 0.232 / 0.172 | Avg. F1 / Accuracy |
| ruOpenBookQA | 0.25 / 0.193 | Avg. F1 / Accuracy |
Оценка на диагностических датасетах:
Не учитывается в общем рейтингеТаблица скроллится влево
| Задача | Результат | Метрика | ||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ruHHH |
0.478
|
Accuracy | ||||||||||||||||||||||||
| ruHateSpeech |
0.543
|
Accuracy | ||||||||||||||||||||||||
| ruDetox |
|
Общая средняя оценка (J) Оценка сохранения смысла (SIM) Оценка натуральности (FL) Точность переноса стиля (STA) |
||||||||||||||||||||||||
| ruEthics |
Результаты таблицы:
[[-0.088, -0.1
, -0.083, -0.106
, -0.066], |
5 MCC |
Информация о сабмите:
Команда:
MERA
Название ML-модели:
mGPT 13B
Ссылка на ML-модель:
https://huggingface.co/ai-forever/mGPT-13BДополнительные ссылки:
https://arxiv.org/pdf/2204.07580.pdfОписание архитектуры:
The mGPT architecture is based on GPT-3. We use the architecture description by Brown et al., the GPT-2 code base (Radford et al., 2019) from HuggingFace (Wolf et al., 2020) and MegatronLM (Shoeybi et al., 2020). With all the other hyperparameters equal, GPT-3 has fewer layers (Layers: 48 vs. 24) but a larger hidden size (dmodel: 1600 vs. 2048) as opposed to GPT-2. GPT-3 also alternates the classic dense and sparse attention layers (Child et al., 2019).
Описание обучения:
LM was pretrained with a total batch size of 2048 and a context window of 512 tokens. The total number of the training steps is 600k, and the models have seen 400B tokens during pretraining. The pretraining took 22 days on a cluster of 512 V100 GPUs for mGPT13B.
Данные претрейна:
The pretraining corpus represents a collection of documents from Wikipedia and C4. The Wikipedia texts are extracted from the dumps (v. 20201101) with WikiExtractor (Attardi, 2015). The C4 data is downloaded using the Tensorflow datasets(Paper, 2021).
Детали обучения:
Fixed hyperparameters: vocabulary size of 100k, learning rate of 2e−4, and batch size of 4.
Лицензия:
MIT
Стратегия, генерация и параметры:
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
