模型:
tinkoff-ai/ruDialoGPT-medium
This generation model is based on sberbank-ai/rugpt3medium_based_on_gpt2 . It's trained on large corpus of dialog data and can be used for buildning generative conversational agents
The model was trained with context size 3
On a private validation set we calculated metrics introduced in this paper :
sensibleness | specificity | SSA | |
---|---|---|---|
tinkoff-ai/ruDialoGPT-small | 0.64 | 0.5 | 0.57 |
tinkoff-ai/ruDialoGPT-medium | 0.78 | 0.69 | 0.735 |
How to use:
import torch from transformers import AutoTokenizer, AutoModelWithLMHead tokenizer = AutoTokenizer.from_pretrained('tinkoff-ai/ruDialoGPT-medium') model = AutoModelWithLMHead.from_pretrained('tinkoff-ai/ruDialoGPT-medium') inputs = tokenizer('@@ПЕРВЫЙ@@ привет @@ВТОРОЙ@@ привет @@ПЕРВЫЙ@@ как дела? @@ВТОРОЙ@@', return_tensors='pt') generated_token_ids = model.generate( **inputs, top_k=10, top_p=0.95, num_beams=3, num_return_sequences=3, do_sample=True, no_repeat_ngram_size=2, temperature=1.2, repetition_penalty=1.2, length_penalty=1.0, eos_token_id=50257, max_new_tokens=40 ) context_with_response = [tokenizer.decode(sample_token_ids) for sample_token_ids in generated_token_ids] context_with_response