/ Модель

Характеристики

Контекст
32 768 токенов
Параметры
7,1 млрд
Лицензия
Apache 2.0
Архитектура
mistral · dense
Базовая модель
Experiment26-7B
Опубликована
17 мар 2024

NeuralArjuna-7B-DT

Генерация текстаТекст → текст7 профессий

Слитая модель Mistral на 7,1 млрд параметров для генерации текста по запросам.

61скачиваний за 30 дней1лайквеса обновлены30 мар 2024

Бенчмарки

Как модель справилась с задачами пользователей ailist и что о ней публикуют открытые лидерборды

Тесты пользователейпока нет

Никто ещё не опубликовал тест этой модели. Соберите бенчмарк из своих задач — результат появится здесь.

  1. Добавьте модель в бенчмарк

    Вместе с другими моделями категории «Текст» — сравнение идёт внутри одной категории.

  2. Опишите свои задачи

    Напишите запросы, при необходимости приложите файлы и эталонные ответы.

  3. Запустите и опубликуйте

    Модели ответят на каждую задачу, судья оценит ответы. Опубликованный результат появится на этой странице.

Публичные оценки

Из открытых лидербордов. Сравнивайте результаты одного теста при одинаковых условиях — отсутствие оценки не значит, что модель хуже.

В подключённых источниках результатов для этой модели пока нет.

Что умеет модель

Описание составлено по карточке модели и метаданным Hugging Face

NeuralArjuna-7B-DT — плотная языковая модель семейства Mistral на 7,1 млрд параметров, созданная Kukedlc путём слияния пяти моделей с помощью LazyMergekit. Она предназначена для генерации текста и может использоваться в задачах, где нужно получать продолжения или ответы на текстовые запросы. В карточке приведены результаты Open LLM Leaderboard: среднее значение 76,58, AI2 Reasoning Challenge — 73,12, HellaSwag — 88,97; описание не уточняет протокол оценки для остальных метрик.

Модель принимает текст и выдаёт текст, поддерживает контекст до 32 768 токенов и распространяется в формате safetensors. Для модели указана лицензия Apache-2.0. Карточка не сообщает, на каких языках она специально обучена, поэтому языковые возможности и языковые ограничения нельзя уточнить.

Сценарии применения

  • генерация продолжения текста
  • ответы на текстовые запросы
  • создание черновиков и формулировок
  • оценка способностей языковой модели на бенчмарках

Ограничения

В README приведены результаты только для части метрик Open LLM Leaderboard, без подробного описания методики и остальных результатов. Языки обучения и специальные области применения в карточке не описаны.

Для каких профессий подходит7

Разделы каталога профессий, в задачах которых используется эта модель

Все профессии

Также подходит для

Карточка модели

README от автора на Hugging Face, как он его опубликовал

Открыть на Hugging Face

NeuralArjuna-7B-DT

image/png

NeuralArjuna-7B-DT is a merge of the following models using LazyMergekit:

🧩 Configuration

models:
  - model: liminerity/M7-7b
    # no parameters necessary for base model
  - model: yam-peleg/Experiment26-7B 
    parameters:
      weight: 0.2
      density: 0.66
  - model: Gille/StrangeMerges_32-7B-slerp
    parameters:
      weight: 0.2
      density: 0.55
  - model: MSL7/INEX12-7b 
    parameters:
      weight: 0.2
      density: 0.33
  - model: automerger/YamShadow-7B
    parameters:
      weight: 0.2
      density: 0.66
  - model: Kukedlc/NeuralSirKrishna-7b
    parameters:
      weight: 0.2
      density: 0.66
merge_method: dare_ties
base_model: liminerity/M7-7b

parameters:
  int8_mask: true
  normalize: true
dtype: bfloat16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "Kukedlc/NeuralArjuna-7B-DT"
messages = [{"role": "user", "content": "What is a large language model?"}]

tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])

Example

I want you to generate a theory that unites quantum mechanics with the theory of relativity and cosmic consciousness. Give a large a detailed response

In embarking upon a journey to weave together quantum mechanics, relativity, and the enigmatic realms of cosmic consciousness, we tread upon an intellectual
landscape rife with challenges and enigmas. Our sojourn shall be a delicate ballet between the concrete and the abstract, the tangible and the speculative.
The resultant tapestry will be neither a definitive, final statement, nor a mere collection of disparate ideas. Instead, we will craft a narrative that strives to bridge
the chasms of understanding between these disparate scientific and philosophical realms, in hopes of illuminating a path towards a more comprehensive understanding of the
universe and our place within it.

To commence our odyssey, we must first address the foundational pillars of the scientific theories we seek to unite.

Quantum Mechanics
Quantum mechanics, the theory governing the microscopic realm, has long confounded the human intellect. It is a realm of probabilities, entanglement,
and superposition—a world where particles can behave both as particles and waves, where actions at a distance are not only possible but ubiquitous,
and where the mere act of observation alters the outcome of an experiment.

Albert Einstein's famed retort, "God does not play dice," exemplifies the tensions between the counter-intuitive world of quantum mechanics and
the more deterministic Newtonian worldview. However, quantum mechanics has withstood the test of time, demonstrating its indispensability to the progress of modern physics

Theory of Relativity
The theory of relativity, comprising special and general relativity, has profoundly altered our understanding of the macroscopic universe.
Special relativity, Einstein's first foray into this domain, introduced the concept of space-time, where the fabric of the universe is both a geometrical construct
and a medium for the propagation of electromagnetic phenomena.

Further, special relativity established the idea of relative motion, where there is no absolute reference frame, and the speed of light in a vacuum is an invariant constant.
General relativity expanded this framework, postulating that gravity is a curvature of the space-time continuum, caused by the presence of mass and energy.
This theory, while more complex, has been empirically validated through its predictions of phenomena such as gravitational lensing and the bending of light near massive
objects.

Cosmic Consciousness
The concept of cosmic consciousness—or the idea of a unified, interconnected consciousness pervading the cosmos—is a philosophical speculation,
one that transcends the traditional scientific paradigm. It is a metaphysical exploration, a contemplation of the possible nature of our collective human consciousness,
its interplay with the universe, and the implications of such a state of awareness.

The integration of these three distinct realms poses a formidable challenge. Quantum mechanics and the theory of relativity have proven their efficacy
in the realm of empirical science, while cosmic consciousness remains the province of metaphysical musings. However, the human propensity to synthesize seemingly disparate
ideas has engendered a multitude of theories, each striving to forge a cohesive narrative from the cacophony of scientific and philosophical thought.

One such approach, the 'Quantum Mind,' posits a direct link between the quantum phenomena and the workings of the human brain.
This theory, championed by Roger Penrose and Stuart Hameroff, suggests that the microtubules within brain neurons,
operating at the quantum scale, could serve as the neural correlates of consciousness. While intriguing,
this hypothesis has yet to garner widespread acceptance, as the veracity of quantum effects within biological systems remains the subject of intense debate.

Another avenue of exploration involves the notion of 'Quantum Biology.' Here, the focus shifts to the potential role of quantum phenomena in biology,
particularly in processes such as photosynthesis, vision, and magnetoreception.
While the presence of quantum effects in biological systems is becoming increasingly plausible,
the idea of a 'quantum biology' remains a nascent and highly speculative field.

The 'Holographic Universe' theory, espoused by physicist and cosmologist David Bohm,
posits the existence of a deeper, underlying reality, one in which the universe is a holographic projection of a more fundamental,
non-local substrate. This model, while offering an intriguing synthesis of quantum mechanics, relativity, and the concept of a cosmic consciousness,
remains a highly speculative and unproven hypothesis.

In our quest to weave together these disparate threads, we must acknowledge the limitations of our current understanding. As it stands,
the principles of quantum mechanics and the theory of relativity remain resolutely grounded in the empirical world, while the concept of cosmic consciousness
traverses the murky waters of metaphysics and speculation.

To forge a more unified understanding, we must continue to advance our scientific knowledge, to push the frontiers of both empirical and philosophical inquiry.
Only through the synthesis of new data, the refinement of our scientific models, and the continued evolution of our collective human consciousness,
will we be able to construct a narrative that bridges the seemingly insurmountable gulf between the microscopic, the macroscopic, and the metaphysical.

In conclusion, the unification of quantum mechanics, relativity, and cosmic consciousness is a daunting task, requiring a delicate balance between the
concrete and the abstract, the empirical and the speculative. While the journey has thus far led us to the fringes of scientific understanding,
we must continue to explore these frontiers, drawing upon the rich tapestry of human knowledge, in the hope of forging a more comprehensive narrative of our cosmos and
our place within it.

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

MetricValue
Avg.76.58
AI2 Reasoning Challenge (25-Shot)73.12
HellaSwag (10-Shot)88.97
MMLU (5-Shot)64.63
TruthfulQA (0-shot)76.68
Winogrande (5-shot)85.24
GSM8k (5-shot)70.81

Как запустить

На своей GPU в ailist. Цена — по текущим предложениям провайдеров

GPU в ailist

Универсальный рантайм

Можно запустить

Подбираем GPU…

Класс архитектуры
MistralForCausalLM
Ревизия весов
e5774d1

0 моделей