Summarization
Transformers
Safetensors
Russian
t5
text2text-generation
fred-t5
russian
research
rahvusarhiiv
Eval Results (legacy)
text-generation-inference
Instructions to use Rahvusarhiiv/ru_summariser with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Rahvusarhiiv/ru_summariser with Transformers:
# Use a pipeline as a high-level helper # Warning: Pipeline type "summarization" is no longer supported in transformers v5. # You must load the model directly (see below) or downgrade to v4.x with: # 'pip install "transformers<5.0.0' from transformers import pipeline pipe = pipeline("summarization", model="Rahvusarhiiv/ru_summariser")# Load model directly from transformers import AutoTokenizer, AutoModelForSeq2SeqLM tokenizer = AutoTokenizer.from_pretrained("Rahvusarhiiv/ru_summariser") model = AutoModelForSeq2SeqLM.from_pretrained("Rahvusarhiiv/ru_summariser", device_map="auto") - Notebooks
- Google Colab
- Kaggle
| { | |
| "release_date": "2026-07-24", | |
| "release_name": "FRED-T5 grounded v3", | |
| "replaces": "mBART-50 Russian summariser", | |
| "previous_hub_tag": "revision-1-mbart", | |
| "base_model": "RussianNLP/FRED-T5-Summarizer", | |
| "architecture": "T5ForConditionalGeneration", | |
| "training": { | |
| "train_rows": 59489, | |
| "validation_rows": 3000, | |
| "max_source_tokens": 768, | |
| "max_target_tokens": 384, | |
| "seed": 20260721, | |
| "steps": 9500, | |
| "epochs": 2.554589924187665, | |
| "best_eval_loss": 1.3249452114105225, | |
| "benchmark_overlap_count": 0, | |
| "synthetic_rows": 0 | |
| }, | |
| "recommended_inference": { | |
| "input_prefix": "summary_length: medium\n\n", | |
| "source_evidence_max_words": 700, | |
| "source_evidence_trigger_words": 760, | |
| "max_input_tokens": 768, | |
| "max_new_tokens": 96, | |
| "min_new_tokens": 0, | |
| "num_beams": 4, | |
| "length_penalty": 0.7, | |
| "no_repeat_ngram_size": 3, | |
| "early_stopping": true, | |
| "truncate_incomplete_sentence": true | |
| }, | |
| "co2_emissions": { | |
| "measured": false, | |
| "reason": "Training energy and emissions were not instrumented." | |
| } | |
| } | |