metadata
tags:
- model_hub_mixin
- pytorch_model_hub_mixin
- gpt-like
- russian
- jokes
language:
- ru
metrics:
- perplexity
model_type: transformer
library_name: pytorch
Russian Jokes GPT-small
“LLM-course HW1”
Модель
Эта модель предназначена для генерации шуток на русском языке. Использует архитектуру трансформера с ALiBi и GQA для позиционного кодирования и внимания.
Использование
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("your_username/my-ru-joke-small")
tokenizer = AutoTokenizer.from_pretrained("your_username/my-ru-joke-small")
prompt = "Встретились два экономиста,"
input_ids = tokenizer(prompt, return_tensors="pt").input_ids
generated_ids = model.generate(input_ids, max_new_tokens=40, temperature=0.9, top_k=40)
print(tokenizer.decode(generated_ids[0], skip_special_tokens=True))
This model has been pushed to the Hub using the [PytorchModelHubMixin](https://huggingface.co/docs/huggingface_hub/package_reference/mixins#huggingface_hub.PyTorchModelHubMixin) integration:
- Code: [More Information Needed]
- Paper: [More Information Needed]
- Docs: [More Information Needed]