Update README.md

README.md

@@ -11,19 +11,41 @@ language:
 
 T-lite-it-1.0 is a model built upon the Qwen 2.5 model family and incorporates both continual pre-training and alignment techniques.
 
-Detailed model card’s coming soon…
-
 ### 📚 Dataset
 
-Detailed model card’s coming soon…
+Pre-training Stage 1:
+100B tokens, consisting of diverse Russian data from Common Crawl, books, code, and proprietary datasets, mixed with re-played English data (English added as it is the primary language of the base model).
+
+Pre-training Stage 2:
+40B tokens, a mix of instruction and pre-training data.
+
+Supervised Fine-Tuning (SFT):
+1B tokens, a mix of diverse instruction data.
+
+Preference Tuning:
+1B tokens, training the model to be helpful.
 
 ## 📊 Benchmarks
 
-Detailed model card’s coming soon…
+| Benchmark                                      | T-lite-it-1.0 | Qwen-2.5-7B-Instruct | GigaChat Pro 1.0.26.15 | RuAdapt-Qwen-7B-Instruct-v1 | gemma-2-9b-it | 
+|------------------------------------------------|:-------------:|:--------------------:|:----------------------:|:---------------------------:|:--------------|
+| [MERA](https://mera.a-ai.ru)                   |   **0.552**   |        0.482         |         0.512          |            0.468            |  0.505 |
+| [MaMuRaMu](https://mera.a-ai.ru/ru/tasks/22)   |   **0.775**   |        0.711         |          0.77          |             0.7             | 0.724  |
+| ruMMLU-PRO                                     |     **0.497**     |        0.481         |           -            |            0.448            |  0.405 |
+| ruGSM8K                                        |     **0.856**     |        0.832         |         0.752          |            0.795            |  0.823 |
+| ruMATH                                         |     **0.679**     |        0.671         |         0.418          |            0.607            |  0.473 |
+ | ruMBPP                                         |     **0.693**     |       0.685         |         0.412          |            0.696            |  0.63 |
+ | [ruCodeEval](https://mera.a-ai.ru/ru/tasks/23) | 0.082 / 0.168 / 0.226 | 0.025 / 0.071 / 0.098 | 0.056 / 0.068 / 0.073 |    0.018 / 0.064 / 0.11     | **0.215 / 0.494 / 0.561** |
+| Arena-Hard-Ru                                  | **64.38** | 54.29 | - |            52.77            | 47.83 | 
+| MT Bench Ru                                    | 7.87 | 7.33 | **8.21** |            7.62             | 7.4 |
+ | Alpaca Eval Ru                                 | **39.61** | 25.61 | 29.83 |            28.43            | 36.87 |
+
+Detailed evaluation results can be found in our [habr post](https://habr.com/ru/companies/tbank/articles/865582/)
 
 
 ## 👨‍💻 Examples of usage
 
+### HF Usage
 
 ```python
 from transformers import AutoTokenizer, AutoModelForCausalLM
@@ -84,4 +106,29 @@ Output:
 Который ведёт к будущему, светлому и новому.  
 Машинное обученье — наш проводник,  
 В этом мире, где технологии царят.
+```
+
+### VLLM Usage
+
+```python
+from transformers import AutoTokenizer
+from vllm import LLM, SamplingParams
+
+model_name = "t-tech/T-lite-it-1.0"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+llm = LLM(model=model_name)
+sampling_params = SamplingParams(temperature=0.3, max_tokens=8192)
+
+prompt = "Напиши стих про машинное обучение"
+messages = [
+    {"role": "system", "content": "Ты T-lite, виртуальный ассистент в Т-Технологии. Твоя задача - быть полезным диалоговым ассистентом."},
+    {"role": "user", "content": prompt}
+]
+
+prompt_token_ids = tokenizer.apply_chat_template(messages, add_generation_prompt=True)
+
+outputs = llm.generate(prompt_token_ids=prompt_token_ids, sampling_params=sampling_params)
+
+generated_text = [output.outputs[0].text for output in outputs]
+print(generated_text)
 ```