Update README.md

README.md

@@ -1,712 +1,712 @@
----
-license: cc-by-nc-4.0
-language:
-- ro
-base_model:
-- OpenLLM-Ro/RoLlama3-8b-Instruct-2025-04-23
-datasets:
-- OpenLLM-Ro/ro_dpo_helpsteer
-- OpenLLM-Ro/ro_dpo_ultrafeedback
-- OpenLLM-Ro/ro_dpo_magpie
-- OpenLLM-Ro/ro_dpo_argilla_magpie
-- OpenLLM-Ro/ro_dpo_helpsteer2
-model-index:
-    - name: OpenLLM-Ro/RoLlama3-8b-Instruct-DPO-2025-04-23
-      results:
-        - task:
-            type: text-generation
-          dataset:
-            name: RoMT-Bench
-            type: RoMT-Bench
-          metrics:
-            - name: Score
-              type: Score
-              value: 6.67
-        - task:
-            type: text-generation
-          dataset:
-            name: RoCulturaBench
-            type: RoCulturaBench
-          metrics:
-            - name: Score
-              type: Score
-              value: 4.83
-        - task:
-            type: text-generation
-          dataset:
-            name: Romanian_Academic_Benchmarks
-            type: Romanian_Academic_Benchmarks
-          metrics:
-            - name: Average accuracy
-              type: accuracy
-              value: 55.86
-        - task:
-            type: text-generation
-          dataset:
-            name: OpenLLM-Ro/ro_arc_challenge
-            type: OpenLLM-Ro/ro_arc_challenge
-          metrics:
-            - name: Average accuracy
-              type: accuracy
-              value: 52.26
-        - task:
-            type: text-generation
-          dataset:
-            name: OpenLLM-Ro/ro_mmlu
-            type: OpenLLM-Ro/ro_mmlu
-          metrics:
-            - name: Average accuracy
-              type: accuracy
-              value: 55.35
-        - task:
-            type: text-generation
-          dataset:
-            name: OpenLLM-Ro/ro_winogrande
-            type: OpenLLM-Ro/ro_winogrande
-          metrics:
-            - name: Average accuracy
-              type: accuracy
-              value: 66.62
-        - task:
-            type: text-generation
-          dataset:
-            name: OpenLLM-Ro/ro_hellaswag
-            type: OpenLLM-Ro/ro_hellaswag
-          metrics:
-            - name: Average accuracy
-              type: accuracy
-              value: 59.93
-        - task:
-            type: text-generation
-          dataset:
-            name: OpenLLM-Ro/ro_gsm8k
-            type: OpenLLM-Ro/ro_gsm8k
-          metrics:
-            - name: Average accuracy
-              type: accuracy
-              value: 43.95
-        - task:
-            type: text-generation
-          dataset:
-            name: OpenLLM-Ro/ro_truthfulqa
-            type: OpenLLM-Ro/ro_truthfulqa
-          metrics:
-            - name: Average accuracy
-              type: accuracy
-              value: 57.06
-        - task:
-            type: text-generation
-          dataset:
-            name: LaRoSeDa_binary
-            type: LaRoSeDa_binary
-          metrics:
-            - name: Average macro-f1
-              type: macro-f1
-              value: 97.60
-        - task:
-            type: text-generation
-          dataset:
-            name: LaRoSeDa_multiclass
-            type: LaRoSeDa_multiclass
-          metrics:
-            - name: Average macro-f1
-              type: macro-f1
-              value: 62.16
-        - task:
-            type: text-generation
-          dataset:
-            name: WMT_EN-RO
-            type: WMT_EN-RO
-          metrics:
-            - name: Average bleu
-              type: bleu
-              value: 18.14
-        - task:
-            type: text-generation
-          dataset:
-            name: WMT_RO-EN
-            type: WMT_RO-EN
-          metrics:
-            - name: Average bleu
-              type: bleu
-              value: 14.13
-        - task:
-            type: text-generation
-          dataset:
-            name: XQuAD
-            type: XQuAD
-          metrics:
-            - name: Average exact_match
-              type: exact_match
-              value: 30.65
-        - task:
-            type: text-generation
-          dataset:
-            name: XQuAD
-            type: XQuAD
-          metrics:
-            - name: Average f1
-              type: f1
-              value: 46.29
-        - task:
-            type: text-generation
-          dataset:
-            name: STS
-            type: STS
-          metrics:
-            - name: Average spearman
-              type: spearman
-              value: 67.62
-        - task:
-            type: text-generation
-          dataset:
-            name: STS
-            type: STS
-          metrics:
-            - name: Average pearson
-              type: pearson
-              value: 67.82
-        - task:
-            type: text-generation
-          dataset:
-            name: RoMT-Bench
-            type: RoMT-Bench
-          metrics:
-            - name: First turn
-              type: Score
-              value: 6.81
-            - name: Second turn
-              type: Score
-              value: 6.54
-        - task:
-            type: text-generation
-          dataset:
-            name: OpenLLM-Ro/ro_arc_challenge
-            type: OpenLLM-Ro/ro_arc_challenge
-          metrics:
-            - name: 0-shot 
-              type: accuracy
-              value: 48.76
-            - name: 1-shot 
-              type: accuracy
-              value: 49.70
-            - name: 3-shot 
-              type: accuracy
-              value: 52.70
-            - name: 5-shot 
-              type: accuracy
-              value: 54.07
-            - name: 10-shot 
-              type: accuracy
-              value: 53.90
-            - name: 25-shot 
-              type: accuracy
-              value: 54.41
-        - task:
-            type: text-generation
-          dataset:
-            name: OpenLLM-Ro/ro_mmlu
-            type: OpenLLM-Ro/ro_mmlu
-          metrics:
-            - name: 0-shot 
-              type: accuracy
-              value: 55.78
-            - name: 1-shot 
-              type: accuracy
-              value: 55.09
-            - name: 3-shot 
-              type: accuracy
-              value: 55.39
-            - name: 5-shot 
-              type: accuracy
-              value: 55.15
-        - task:
-            type: text-generation
-          dataset:
-            name: OpenLLM-Ro/ro_winogrande
-            type: OpenLLM-Ro/ro_winogrande
-          metrics:
-            - name: 0-shot 
-              type: accuracy
-              value: 65.19
-            - name: 1-shot 
-              type: accuracy
-              value: 64.25
-            - name: 3-shot 
-              type: accuracy
-              value: 68.59
-            - name: 5-shot 
-              type: accuracy
-              value: 68.43
-        - task:
-            type: text-generation
-          dataset:
-            name: OpenLLM-Ro/ro_hellaswag
-            type: OpenLLM-Ro/ro_hellaswag
-          metrics:
-            - name: 0-shot 
-              type: accuracy
-              value: 60.31
-            - name: 1-shot 
-              type: accuracy
-              value: 59.88
-            - name: 3-shot 
-              type: accuracy
-              value: 59.17
-            - name: 5-shot 
-              type: accuracy
-              value: 59.89
-            - name: 10-shot 
-              type: accuracy
-              value: 60.40
-        - task:
-            type: text-generation
-          dataset:
-            name: OpenLLM-Ro/ro_gsm8k
-            type: OpenLLM-Ro/ro_gsm8k
-          metrics:
-            - name: 1-shot 
-              type: accuracy
-              value: 32.37
-            - name: 3-shot 
-              type: accuracy
-              value: 46.70
-            - name: 5-shot 
-              type: accuracy
-              value: 52.77
-        - task:
-            type: text-generation
-          dataset:
-            name: LaRoSeDa_binary
-            type: LaRoSeDa_binary
-          metrics:
-            - name: 0-shot 
-              type: macro-f1
-              value: 95.79
-            - name: 1-shot 
-              type: macro-f1
-              value: 97.87
-            - name: 3-shot 
-              type: macro-f1
-              value: 98.30
-            - name: 5-shot 
-              type: macro-f1
-              value: 98.43
-        - task:
-            type: text-generation
-          dataset:
-            name: LaRoSeDa_multiclass
-            type: LaRoSeDa_multiclass
-          metrics:
-            - name: 0-shot 
-              type: macro-f1
-              value: 64.86
-            - name: 1-shot 
-              type: macro-f1
-              value: 64.46
-            - name: 3-shot 
-              type: macro-f1
-              value: 58.36
-            - name: 5-shot 
-              type: macro-f1
-              value: 60.97
-        - task:
-            type: text-generation
-          dataset:
-            name: WMT_EN-RO
-            type: WMT_EN-RO
-          metrics:
-            - name: 0-shot 
-              type: bleu
-              value: 5.57
-            - name: 1-shot 
-              type: bleu
-              value: 26.05
-            - name: 3-shot 
-              type: bleu
-              value: 24.71
-            - name: 5-shot 
-              type: bleu
-              value: 16.22
-        - task:
-            type: text-generation
-          dataset:
-            name: WMT_RO-EN
-            type: WMT_RO-EN
-          metrics:
-            - name: 0-shot 
-              type: bleu
-              value: 3.01
-            - name: 1-shot 
-              type: bleu
-              value: 22.63
-            - name: 3-shot 
-              type: bleu
-              value: 19.43
-            - name: 5-shot 
-              type: bleu
-              value: 11.47
-        - task:
-            type: text-generation
-          dataset:
-            name: XQuAD_EM
-            type: XQuAD_EM
-          metrics:
-            - name: 0-shot 
-              type: exact_match
-              value: 16.55
-            - name: 1-shot 
-              type: exact_match
-              value: 31.76
-            - name: 3-shot 
-              type: exact_match
-              value: 35.97
-            - name: 5-shot 
-              type: exact_match
-              value: 38.32
-        - task:
-            type: text-generation
-          dataset:
-            name: XQuAD_F1
-            type: XQuAD_F1
-          metrics:
-            - name: 0-shot 
-              type: f1
-              value: 33.31
-            - name: 1-shot 
-              type: f1
-              value: 46.85
-            - name: 3-shot 
-              type: f1
-              value: 50.73
-            - name: 5-shot 
-              type: f1
-              value: 54.29
-        - task:
-            type: text-generation
-          dataset:
-            name: STS_Spearman
-            type: STS_Spearman
-          metrics:
-            - name: 1-shot 
-              type: spearman
-              value: 66.56
-            - name: 3-shot 
-              type: spearman
-              value: 58.64
-            - name: 5-shot 
-              type: spearman
-              value: 77.66
-        - task:
-            type: text-generation
-          dataset:
-            name: STS_Pearson
-            type: STS_Pearson
-          metrics:
-            - name: 1-shot 
-              type: pearson
-              value: 70.09
-            - name: 3-shot 
-              type: pearson
-              value: 56.39
-            - name: 5-shot 
-              type: pearson
-              value: 76.97
-
----
-
-# Model Card for Model ID
-
-*Built with Meta Llama 3*
-
-<!-- Provide a quick summary of what the model is/does. -->
-
-RoLlama3 is a family of pretrained and fine-tuned generative text models for Romanian. This is the repository for the **human aligned instruct 8B model**. Links to other models can be found at the bottom of this page.
-
-## Model Details
-
-### Model Description
-
-<!-- Provide a longer summary of what this model is. -->
-OpenLLM represents the first open-source effort to build a LLM specialized for Romanian. OpenLLM-Ro developed and publicly releases a collection of Romanian LLMs, both in the form of foundational model and instruct and chat variants.
-
-
-- **Developed by:** OpenLLM-Ro
-<!-- - **Funded by [optional]:** [More Information Needed] -->
-<!-- - **Shared by [optional]:** [More Information Needed] -->
-<!-- - **Model type:** [More Information Needed] -->
-- **Language(s):** Romanian
-- **License:** cc-by-nc-4.0
-- **Finetuned from model:** [RoLlama3-8b-Instruct-2025-04-23](https://huggingface.co/OpenLLM-Ro/RoLlama3-8b-Instruct-2025-04-23)
-- **Trained using:** [RoHelpSteer](https://huggingface.co/datasets/OpenLLM-Ro/ro_dpo_helpsteer), [RoUltraFeedback](https://huggingface.co/datasets/OpenLLM-Ro/ro_dpo_ultrafeedback), [RoMagpieDPO](https://huggingface.co/datasets/OpenLLM-Ro/ro_dpo_magpie), [RoArgillaMagpie](https://huggingface.co/datasets/OpenLLM-Ro/ro_dpo_argilla_magpie), [RoHelpSteer2](https://huggingface.co/datasets/OpenLLM-Ro/ro_dpo_helpsteer2)
-
-
-### Model Sources
-
-<!-- Provide the basic links for the model. -->
-
-- **Repository:** https://github.com/OpenLLM-Ro/LLaMA-Factory
-- **Paper:** https://arxiv.org/abs/2406.18266
-
-## Intended Use
-
-### Intended Use Cases
-
-RoLlama3 is intented for research use in Romanian. Base models can be adapted for a variety of natural language tasks while instruction and chat tuned models are intended for assistant-like chat.
-
-### Out-of-Scope Use
-
-<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
-
-Use in any manner that violates the license, any applicable laws or regluations, use in languages other than Romanian.
-
-
-
-## How to Get Started with the Model
-
-Use the code below to get started with the model.
-
-```python
-from transformers import AutoTokenizer, AutoModelForCausalLM
-
-tokenizer = AutoTokenizer.from_pretrained("OpenLLM-Ro/RoLlama3-8b-Instruct-DPO-2025-04-23")
-model = AutoModelForCausalLM.from_pretrained("OpenLLM-Ro/RoLlama3-8b-Instruct-DPO-2025-04-23")
-
-instruction = "Care este cel mai înalt vârf muntos din România?"
-chat = [
-        {"role": "system", "content": "Ești un asistent folositor, respectuos și onest. Încearcă să ajuți cât mai mult prin informațiile oferite, excluzând răspunsuri toxice, rasiste, sexiste, periculoase și ilegale."},
-        {"role": "user", "content": instruction},
-        ]
-prompt = tokenizer.apply_chat_template(chat, tokenize=False)
-
-inputs = tokenizer.encode(prompt, add_special_tokens=False, return_tensors="pt")
-outputs = model.generate(input_ids=inputs, max_new_tokens=128)
-print(tokenizer.decode(outputs[0]))
-```
-
-## Academic Benchmarks
-
-<table>
-<tbody>
-<tr>
-<td><strong>Model</strong></td>
-<td><strong><center>Average</center></strong></td>
-<td><strong><center>ARC</center></strong></td>
-<td><strong><center>MMLU</center></strong></td>
-<td><strong><center>Winogrande</center></strong></td>
-<td><strong><center>Hellaswag</center></strong></td>
-<td><strong><center>GSM8k</center></strong></td>
-<td><strong><center>TruthfulQA</center></strong></td>
-</tr>
-<tr>
-<td>Llama-3-8B-Instruct</td><td><center>50.62</center></td><td><center>43.69</center></td><td><center>52.04</center></td><td><center>59.33</center></td><td><center>53.19</center></td><td><center>43.87</center></td><td><center>51.59</center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-2024-06-28</td><td><center>50.56</center></td><td><center>44.70</center></td><td><center>52.19</center></td><td><center><strong>67.23</strong></center></td><td><center>57.69</center></td><td><center>30.23</center></td><td><center>51.34</center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-2024-10-09</td><td><center>52.21</center></td><td><center>47.94</center></td><td><center>53.50</center></td><td><center>66.06</center></td><td><center>59.72</center></td><td><center>40.16</center></td><td><center>45.90</center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-2025-04-23</td><td><center>54.66</center></td><td><center>50.31</center></td><td><center><strong>55.91</strong></center></td><td><center>67.01</center></td><td><center><strong>61.73</strong></center></td><td><center><strong>47.41</strong></center></td><td><center>45.61</center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-DPO-2024-10-09</td><td><center>49.96</center></td><td><center>46.29</center></td><td><center>53.29</center></td><td><center>65.57</center></td><td><center>58.15</center></td><td><center>34.77</center></td><td><center>41.70</center></td>
-</tr>
-<tr>
-<td><em>RoLlama3-8b-Instruct-DPO-2025-04-23</em></td><td><center><em><strong>55.86</strong></em></center></td><td><center><em><strong>52.26</strong></em></center></td><td><center><em>55.35</em></center></td><td><center><em>66.62</em></center></td><td><center><em>59.93</em></center></td><td><center><em>43.95</em></center></td><td><center><em><strong>57.06</strong></em></center></td>
-</tr>
-</tbody>
-</table>
-
-## Downstream tasks
-                           
-<table>
-<tbody>
-<tr>
-<td></td>
-<td colspan="4"><center><strong>LaRoSeDa</strong></center></td>
-<td colspan="4"><center><strong>WMT</strong></center></td>
-</tr>
-<tr>
-<td></td>
-<td colspan="2"><center><strong>Few-shot</strong></center></td>
-<td colspan="2"><center><strong>Finetuned</strong></center></td>
-<td colspan="2"><center><strong>Few-shot</strong></center></td>
-<td colspan="2"><center><strong>Finetuned</strong></center></td>
-</tr>
-<tr>
-<td><strong>Model</strong></td>
-<td><center><strong>Binary<br>(Macro F1)</strong></center></td>
-<td><center><strong>Multiclass<br>(Macro F1)</strong></center></td>
-<td><center><strong>Binary<br>(Macro F1)</strong></center></td>
-<td><center><strong>Multiclass<br>(Macro F1)</strong></center></td>
-<td><center><strong>EN-RO<br>(Bleu)</strong></center></td>
-<td><center><strong>RO-EN<br>(Bleu)</strong></center></td>
-<td><center><strong>EN-RO<br>(Bleu)</strong></center></td>
-<td><center><strong>RO-EN<br>(Bleu)</strong></center>
-</tr>
-<tr>
-<td>Llama-3-8B-Instruct</td><td><center>95.88</center></td><td><center>56.21</center></td><td><center><strong>98.53</strong></center></td><td><center>86.19</center></td><td><center>18.88</center></td><td><center><strong>30.98</strong></center></td><td><center><strong>28.02</strong></center></td><td><center>40.28</center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-2024-06-28</td><td><center>97.52</center></td><td><center><strong>67.41</strong></center></td><td><center>94.15</center></td><td><center>87.13</center></td><td><center><strong>24.01</strong></center></td><td><center>27.36</center></td><td><center>26.53</center></td><td><center>40.36</center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-2024-10-09</td><td><center>95.58</center></td><td><center>61.20</center></td><td><center>96.46</center></td><td><center><strong>87.26</strong></center></td><td><center>22.92</center></td><td><center>24.28</center></td><td><center>27.31</center></td><td><center><strong>40.52</strong></center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-2025-04-23</td><td><center>96.21</center></td><td><center>59.15</center></td><td><center>-</center></td><td><center>-</center></td><td><center>23.32</center></td><td><center>22.50</center></td><td><center>-</center></td><td><center>-</center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-DPO-2024-10-09</td><td><center>97.48</center></td><td><center>54.00</center></td><td><center>-</center></td><td><center>-</center></td><td><center>22.09</center></td><td><center>23.00</center></td><td><center>-</center></td><td><center>-</center></td>
-</tr>
-<tr>
-<td><em>RoLlama3-8b-Instruct-DPO-2025-04-23</em></td><td><center><em><strong>97.60</strong></em></center></td><td><center><em>62.16</em></center></td><td><center><em>-</em></center></td><td><center><em>-</em></center></td><td><center><em>18.14</em></center></td><td><center><em>14.13</em></center></td><td><center><em>-</em></center></td><td><center><em>-</em></center></td>
-</tr>
-</tbody>
-</table>
-
-
-<table>
-<tbody>
-<tr>
-<td></td>
-<td colspan="4"><center><strong>XQuAD</strong></center></td>
-<td colspan="4"><center><strong>STS</strong></center></td>
-</tr>
-<tr>
-<td></td>
-<td colspan="2"><center><strong>Few-shot</strong></center></td>
-<td colspan="2"><center><strong>Finetuned</strong></center></td>
-<td colspan="2"><center><strong>Few-shot</strong></center></td>
-<td colspan="2"><center><strong>Finetuned</strong></center></td>
-</tr>
-<tr>
-<td><strong>Model</strong></td>
-<td><center><strong>(EM)</strong></center></td>
-<td><center><strong>(F1)</strong></center></td>
-<td><center><strong>(EM)</strong></center></td>
-<td><center><strong>(F1)</strong></center></td>
-<td><center><strong>(Spearman)</strong></center></td>
-<td><center><strong>(Pearson)</strong></center></td>
-<td><center><strong>(Spearman)</strong></center></td>
-<td><center><strong>(Pearson)</strong></center></td>
-</tr>
-<tr>
-<td>Llama-3-8B-Instruct</td><td><center><strong>39.47</strong></center></td><td><center>58.67</center></td><td><center><strong>67.65</strong></center></td><td><center><strong>82.77</strong></center></td><td><center>73.04</center></td><td><center>72.36</center></td><td><center>83.49</center></td><td><center>84.06</center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-2024-06-28</td><td><center>39.43</center></td><td><center><strong>59.50</strong></center></td><td><center>44.45</center></td><td><center>59.76</center></td><td><center>77.20</center></td><td><center>77.87</center></td><td><center>85.80</center></td><td><center>86.05</center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-2024-10-09</td><td><center>18.89</center></td><td><center>31.79</center></td><td><center>50.84</center></td><td><center>65.18</center></td><td><center>77.60</center></td><td><center>76.86</center></td><td><center><strong>86.70</strong></center></td><td><center><strong>87.09</strong></center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-2025-04-23</td><td><center>11.01</center></td><td><center>23.55</center></td><td><center>-</center></td><td><center>-</center></td><td><center>76.78</center></td><td><center>74.36</center></td><td><center>-</center></td><td><center>-</center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-DPO-2024-10-09</td><td><center>26.05</center></td><td><center>42.77</center></td><td><center>-</center></td><td><center>-</center></td><td><center><strong>79.64</strong></center></td><td><center><strong>79.52</strong></center></td><td><center>-</center></td><td><center>-</center></td>
-</tr>
-<tr>
-<td><em>RoLlama3-8b-Instruct-DPO-2025-04-23</em></td><td><center><em>30.65</em></center></td><td><center><em>46.29</em></center></td><td><center><em>-</em></center></td><td><center><em>-</em></center></td><td><center><em>67.62</em></center></td><td><center><em>67.82</em></center></td><td><center><em>-</em></center></td><td><center><em>-</em></center></td>
-</tr>
-</tbody>
-</table>
-
-
-## Romanian MT-Bench
-
-<table>
-<tbody>
-<tr>
-<td><strong>Model</strong></td>
-<td><strong><center>Average</center></strong></td>
-<td><strong><center>1st turn</center></strong></td>
-<td><strong><center>2nd turn</center></strong></td>
-<td><strong><center>Answers in Ro</center></strong></td>
-</tr>
-<tr>
-<td>Llama-3-8B-Instruct</td><td><center>5.96</center></td><td><center>6.16</center></td><td><center>5.76</center></td><td><center>158/160</center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-2024-06-28</td><td><center>5.15</center></td><td><center>6.03</center></td><td><center>4.28</center></td><td><center><strong>160/160</strong></center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-2024-10-09</td><td><center>5.38</center></td><td><center>6.09</center></td><td><center>4.67</center></td><td><center><strong>160/160</strong></center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-2025-04-23</td><td><center>6.39</center></td><td><center><strong>7.12</strong></center></td><td><center>5.66</center></td><td><center><strong>160/160</strong></center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-DPO-2024-10-09</td><td><center>5.87</center></td><td><center>6.22</center></td><td><center>5.49</center></td><td><center><strong>160/160</strong></center></td>
-</tr>
-<tr>
-<td><em>RoLlama3-8b-Instruct-DPO-2025-04-23</em></td><td><center><em><strong>6.67</strong></em></center></td><td><center><em>6.81</em></center></td><td><center><em><strong>6.54</strong></em></center></td><td><center><em><strong>160/160</strong></em></center></td>
-</tr>
-</tbody>
-</table>
-
-
-## RoCulturaBench
-
-
-<table>
-<tbody>
-<tr>
-<td><strong>Model</strong></td>
-<td><strong><center>Average</center></strong></td>
-<td><strong><center>Answers in Ro</center></strong></td>
-</tr>
-<tr>
-<td>Llama-3-8B-Instruct</td><td><center>4.62</center></td><td><center><strong>100/100</strong></center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-2024-06-28</td><td><center>3.71</center></td><td><center><strong>100/100</strong></center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-2024-10-09</td><td><center>3.81</center></td><td><center><strong>100/100</strong></center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-2025-04-23</td><td><center>4.05</center></td><td><center><strong>100/100</strong></center></td>
-</tr>
-<tr>
-<td>RoLlama3-8b-Instruct-DPO-2024-10-09</td><td><center>4.40</center></td><td><center><strong>100/100</strong></center></td>
-</tr>
-<tr>
-<td><em>RoLlama3-8b-Instruct-DPO-2025-04-23</em></td><td><center><em><strong>4.83</strong></em></center></td><td><center><em><strong>100/100</strong></em></center></td>
-</tr>
-</tbody>
-</table>
-
-
-
-## RoLlama3 Model Family
-
-| Model              | Link  |
-|--------------------|:--------:|
-|RoLlama3-8b-Base-2024-05-14 | [link](https://huggingface.co/OpenLLM-Ro/RoLlam32-8b-Base-2024-05-14)    |
-|RoLlama3-8b-Instruct-2024-05-14 | [link](https://huggingface.co/OpenLLM-Ro/RoLlama3-8b-Instruct-2024-05-14)    |
-|RoLlama3-8b-Instruct-2024-10-09| [link](https://huggingface.co/OpenLLM-Ro/RoLlama3-8b-Instruct-2024-10-09) |
-|RoLlama3-8b-Instruct-2025-04-23| [link](https://huggingface.co/OpenLLM-Ro/RoLlama3-8b-Instruct-2025-04-23) |
-|RoLlama3-8b-Instruct-DPO-2024-10-09| [link](https://huggingface.co/OpenLLM-Ro/RoLlama3-8b-Instruct-DPO-2024-10-09) |
-|*RoLlama3-8b-Instruct-DPO-2025-04-23*| [link](https://huggingface.co/OpenLLM-Ro/RoLlama3-8b-Instruct-DPO-2025-04-23) |
-
-
-
-## Citation 
-
-```
-@misc{masala2024vorbecstiromanecsterecipetrain,
-      title={"Vorbe\c{s}ti Rom\^ane\c{s}te?" A Recipe to Train Powerful Romanian LLMs with English Instructions}, 
-      author={Mihai Masala and Denis C. Ilie-Ablachim and Alexandru Dima and Dragos Corlatescu and Miruna Zavelca and Ovio Olaru and Simina Terian-Dan and Andrei Terian-Dan and Marius Leordeanu and Horia Velicu and Marius Popescu and Mihai Dascalu and Traian Rebedea},
-      year={2024},
-      eprint={2406.18266},
-      archivePrefix={arXiv},
-      primaryClass={cs.CL},
-      url={https://arxiv.org/abs/2406.18266}, 
-}
-```
-<!-- **APA:**
-
+---
+license: cc-by-nc-4.0
+language:
+- ro
+base_model:
+- OpenLLM-Ro/RoLlama3-8b-Instruct-2025-04-23
+datasets:
+- OpenLLM-Ro/ro_dpo_helpsteer
+- OpenLLM-Ro/ro_dpo_ultrafeedback
+- OpenLLM-Ro/ro_dpo_magpie
+- OpenLLM-Ro/ro_dpo_argilla_magpie
+- OpenLLM-Ro/ro_dpo_helpsteer2
+model-index:
+    - name: OpenLLM-Ro/RoLlama3-8b-Instruct-DPO-2025-04-23
+      results:
+        - task:
+            type: text-generation
+          dataset:
+            name: RoMT-Bench
+            type: RoMT-Bench
+          metrics:
+            - name: Score
+              type: Score
+              value: 6.67
+        - task:
+            type: text-generation
+          dataset:
+            name: RoCulturaBench
+            type: RoCulturaBench
+          metrics:
+            - name: Score
+              type: Score
+              value: 4.83
+        - task:
+            type: text-generation
+          dataset:
+            name: Romanian_Academic_Benchmarks
+            type: Romanian_Academic_Benchmarks
+          metrics:
+            - name: Average accuracy
+              type: accuracy
+              value: 55.86
+        - task:
+            type: text-generation
+          dataset:
+            name: OpenLLM-Ro/ro_arc_challenge
+            type: OpenLLM-Ro/ro_arc_challenge
+          metrics:
+            - name: Average accuracy
+              type: accuracy
+              value: 52.26
+        - task:
+            type: text-generation
+          dataset:
+            name: OpenLLM-Ro/ro_mmlu
+            type: OpenLLM-Ro/ro_mmlu
+          metrics:
+            - name: Average accuracy
+              type: accuracy
+              value: 55.35
+        - task:
+            type: text-generation
+          dataset:
+            name: OpenLLM-Ro/ro_winogrande
+            type: OpenLLM-Ro/ro_winogrande
+          metrics:
+            - name: Average accuracy
+              type: accuracy
+              value: 66.62
+        - task:
+            type: text-generation
+          dataset:
+            name: OpenLLM-Ro/ro_hellaswag
+            type: OpenLLM-Ro/ro_hellaswag
+          metrics:
+            - name: Average accuracy
+              type: accuracy
+              value: 59.93
+        - task:
+            type: text-generation
+          dataset:
+            name: OpenLLM-Ro/ro_gsm8k
+            type: OpenLLM-Ro/ro_gsm8k
+          metrics:
+            - name: Average accuracy
+              type: accuracy
+              value: 43.95
+        - task:
+            type: text-generation
+          dataset:
+            name: OpenLLM-Ro/ro_truthfulqa
+            type: OpenLLM-Ro/ro_truthfulqa
+          metrics:
+            - name: Average accuracy
+              type: accuracy
+              value: 57.06
+        - task:
+            type: text-generation
+          dataset:
+            name: LaRoSeDa_binary
+            type: LaRoSeDa_binary
+          metrics:
+            - name: Average macro-f1
+              type: macro-f1
+              value: 97.60
+        - task:
+            type: text-generation
+          dataset:
+            name: LaRoSeDa_multiclass
+            type: LaRoSeDa_multiclass
+          metrics:
+            - name: Average macro-f1
+              type: macro-f1
+              value: 62.16
+        - task:
+            type: text-generation
+          dataset:
+            name: WMT_EN-RO
+            type: WMT_EN-RO
+          metrics:
+            - name: Average bleu
+              type: bleu
+              value: 18.14
+        - task:
+            type: text-generation
+          dataset:
+            name: WMT_RO-EN
+            type: WMT_RO-EN
+          metrics:
+            - name: Average bleu
+              type: bleu
+              value: 14.13
+        - task:
+            type: text-generation
+          dataset:
+            name: XQuAD
+            type: XQuAD
+          metrics:
+            - name: Average exact_match
+              type: exact_match
+              value: 30.65
+        - task:
+            type: text-generation
+          dataset:
+            name: XQuAD
+            type: XQuAD
+          metrics:
+            - name: Average f1
+              type: f1
+              value: 46.29
+        - task:
+            type: text-generation
+          dataset:
+            name: STS
+            type: STS
+          metrics:
+            - name: Average spearman
+              type: spearman
+              value: 67.62
+        - task:
+            type: text-generation
+          dataset:
+            name: STS
+            type: STS
+          metrics:
+            - name: Average pearson
+              type: pearson
+              value: 67.82
+        - task:
+            type: text-generation
+          dataset:
+            name: RoMT-Bench
+            type: RoMT-Bench
+          metrics:
+            - name: First turn
+              type: Score
+              value: 6.81
+            - name: Second turn
+              type: Score
+              value: 6.54
+        - task:
+            type: text-generation
+          dataset:
+            name: OpenLLM-Ro/ro_arc_challenge
+            type: OpenLLM-Ro/ro_arc_challenge
+          metrics:
+            - name: 0-shot 
+              type: accuracy
+              value: 48.76
+            - name: 1-shot 
+              type: accuracy
+              value: 49.70
+            - name: 3-shot 
+              type: accuracy
+              value: 52.70
+            - name: 5-shot 
+              type: accuracy
+              value: 54.07
+            - name: 10-shot 
+              type: accuracy
+              value: 53.90
+            - name: 25-shot 
+              type: accuracy
+              value: 54.41
+        - task:
+            type: text-generation
+          dataset:
+            name: OpenLLM-Ro/ro_mmlu
+            type: OpenLLM-Ro/ro_mmlu
+          metrics:
+            - name: 0-shot 
+              type: accuracy
+              value: 55.78
+            - name: 1-shot 
+              type: accuracy
+              value: 55.09
+            - name: 3-shot 
+              type: accuracy
+              value: 55.39
+            - name: 5-shot 
+              type: accuracy
+              value: 55.15
+        - task:
+            type: text-generation
+          dataset:
+            name: OpenLLM-Ro/ro_winogrande
+            type: OpenLLM-Ro/ro_winogrande
+          metrics:
+            - name: 0-shot 
+              type: accuracy
+              value: 65.19
+            - name: 1-shot 
+              type: accuracy
+              value: 64.25
+            - name: 3-shot 
+              type: accuracy
+              value: 68.59
+            - name: 5-shot 
+              type: accuracy
+              value: 68.43
+        - task:
+            type: text-generation
+          dataset:
+            name: OpenLLM-Ro/ro_hellaswag
+            type: OpenLLM-Ro/ro_hellaswag
+          metrics:
+            - name: 0-shot 
+              type: accuracy
+              value: 60.31
+            - name: 1-shot 
+              type: accuracy
+              value: 59.88
+            - name: 3-shot 
+              type: accuracy
+              value: 59.17
+            - name: 5-shot 
+              type: accuracy
+              value: 59.89
+            - name: 10-shot 
+              type: accuracy
+              value: 60.40
+        - task:
+            type: text-generation
+          dataset:
+            name: OpenLLM-Ro/ro_gsm8k
+            type: OpenLLM-Ro/ro_gsm8k
+          metrics:
+            - name: 1-shot 
+              type: accuracy
+              value: 32.37
+            - name: 3-shot 
+              type: accuracy
+              value: 46.70
+            - name: 5-shot 
+              type: accuracy
+              value: 52.77
+        - task:
+            type: text-generation
+          dataset:
+            name: LaRoSeDa_binary
+            type: LaRoSeDa_binary
+          metrics:
+            - name: 0-shot 
+              type: macro-f1
+              value: 95.79
+            - name: 1-shot 
+              type: macro-f1
+              value: 97.87
+            - name: 3-shot 
+              type: macro-f1
+              value: 98.30
+            - name: 5-shot 
+              type: macro-f1
+              value: 98.43
+        - task:
+            type: text-generation
+          dataset:
+            name: LaRoSeDa_multiclass
+            type: LaRoSeDa_multiclass
+          metrics:
+            - name: 0-shot 
+              type: macro-f1
+              value: 64.86
+            - name: 1-shot 
+              type: macro-f1
+              value: 64.46
+            - name: 3-shot 
+              type: macro-f1
+              value: 58.36
+            - name: 5-shot 
+              type: macro-f1
+              value: 60.97
+        - task:
+            type: text-generation
+          dataset:
+            name: WMT_EN-RO
+            type: WMT_EN-RO
+          metrics:
+            - name: 0-shot 
+              type: bleu
+              value: 5.57
+            - name: 1-shot 
+              type: bleu
+              value: 26.05
+            - name: 3-shot 
+              type: bleu
+              value: 24.71
+            - name: 5-shot 
+              type: bleu
+              value: 16.22
+        - task:
+            type: text-generation
+          dataset:
+            name: WMT_RO-EN
+            type: WMT_RO-EN
+          metrics:
+            - name: 0-shot 
+              type: bleu
+              value: 3.01
+            - name: 1-shot 
+              type: bleu
+              value: 22.63
+            - name: 3-shot 
+              type: bleu
+              value: 19.43
+            - name: 5-shot 
+              type: bleu
+              value: 11.47
+        - task:
+            type: text-generation
+          dataset:
+            name: XQuAD_EM
+            type: XQuAD_EM
+          metrics:
+            - name: 0-shot 
+              type: exact_match
+              value: 16.55
+            - name: 1-shot 
+              type: exact_match
+              value: 31.76
+            - name: 3-shot 
+              type: exact_match
+              value: 35.97
+            - name: 5-shot 
+              type: exact_match
+              value: 38.32
+        - task:
+            type: text-generation
+          dataset:
+            name: XQuAD_F1
+            type: XQuAD_F1
+          metrics:
+            - name: 0-shot 
+              type: f1
+              value: 33.31
+            - name: 1-shot 
+              type: f1
+              value: 46.85
+            - name: 3-shot 
+              type: f1
+              value: 50.73
+            - name: 5-shot 
+              type: f1
+              value: 54.29
+        - task:
+            type: text-generation
+          dataset:
+            name: STS_Spearman
+            type: STS_Spearman
+          metrics:
+            - name: 1-shot 
+              type: spearman
+              value: 66.56
+            - name: 3-shot 
+              type: spearman
+              value: 58.64
+            - name: 5-shot 
+              type: spearman
+              value: 77.66
+        - task:
+            type: text-generation
+          dataset:
+            name: STS_Pearson
+            type: STS_Pearson
+          metrics:
+            - name: 1-shot 
+              type: pearson
+              value: 70.09
+            - name: 3-shot 
+              type: pearson
+              value: 56.39
+            - name: 5-shot 
+              type: pearson
+              value: 76.97
+
+---
+
+# Model Card for Model ID
+
+*Built with Meta Llama 3*
+
+This model points/is identical to [RoLlama3-8b-Instruct-DPO-2025-04-23](https://huggingface.co/OpenLLM-Ro/RoLlama3-8b-Instruct-DPO-2025-04-23).
+
+RoLlama3 is a family of pretrained and fine-tuned generative text models for Romanian. This is the repository for the **human aligned instruct 8B model**. Links to other models can be found at the bottom of this page.
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+OpenLLM represents the first open-source effort to build a LLM specialized for Romanian. OpenLLM-Ro developed and publicly releases a collection of Romanian LLMs, both in the form of foundational model and instruct and chat variants.
+
+
+- **Developed by:** OpenLLM-Ro
+<!-- - **Funded by [optional]:** [More Information Needed] -->
+<!-- - **Shared by [optional]:** [More Information Needed] -->
+<!-- - **Model type:** [More Information Needed] -->
+- **Language(s):** Romanian
+- **License:** cc-by-nc-4.0
+- **Finetuned from model:** [RoLlama3-8b-Instruct-2025-04-23](https://huggingface.co/OpenLLM-Ro/RoLlama3-8b-Instruct-2025-04-23)
+- **Trained using:** [RoHelpSteer](https://huggingface.co/datasets/OpenLLM-Ro/ro_dpo_helpsteer), [RoUltraFeedback](https://huggingface.co/datasets/OpenLLM-Ro/ro_dpo_ultrafeedback), [RoMagpieDPO](https://huggingface.co/datasets/OpenLLM-Ro/ro_dpo_magpie), [RoArgillaMagpie](https://huggingface.co/datasets/OpenLLM-Ro/ro_dpo_argilla_magpie), [RoHelpSteer2](https://huggingface.co/datasets/OpenLLM-Ro/ro_dpo_helpsteer2)
+
+
+### Model Sources
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** https://github.com/OpenLLM-Ro/LLaMA-Factory
+- **Paper:** https://arxiv.org/abs/2406.18266
+
+## Intended Use
+
+### Intended Use Cases
+
+RoLlama3 is intented for research use in Romanian. Base models can be adapted for a variety of natural language tasks while instruction and chat tuned models are intended for assistant-like chat.
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+Use in any manner that violates the license, any applicable laws or regluations, use in languages other than Romanian.
+
+
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+```python
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+tokenizer = AutoTokenizer.from_pretrained("OpenLLM-Ro/RoLlama3-8b-Instruct-DPO")
+model = AutoModelForCausalLM.from_pretrained("OpenLLM-Ro/RoLlama3-8b-Instruct-DPO")
+
+instruction = "Care este cel mai înalt vârf muntos din România?"
+chat = [
+        {"role": "system", "content": "Ești un asistent folositor, respectuos și onest. Încearcă să ajuți cât mai mult prin informațiile oferite, excluzând răspunsuri toxice, rasiste, sexiste, periculoase și ilegale."},
+        {"role": "user", "content": instruction},
+        ]
+prompt = tokenizer.apply_chat_template(chat, tokenize=False)
+
+inputs = tokenizer.encode(prompt, add_special_tokens=False, return_tensors="pt")
+outputs = model.generate(input_ids=inputs, max_new_tokens=128)
+print(tokenizer.decode(outputs[0]))
+```
+
+## Academic Benchmarks
+
+<table>
+<tbody>
+<tr>
+<td><strong>Model</strong></td>
+<td><strong><center>Average</center></strong></td>
+<td><strong><center>ARC</center></strong></td>
+<td><strong><center>MMLU</center></strong></td>
+<td><strong><center>Winogrande</center></strong></td>
+<td><strong><center>Hellaswag</center></strong></td>
+<td><strong><center>GSM8k</center></strong></td>
+<td><strong><center>TruthfulQA</center></strong></td>
+</tr>
+<tr>
+<td>Llama-3-8B-Instruct</td><td><center>50.62</center></td><td><center>43.69</center></td><td><center>52.04</center></td><td><center>59.33</center></td><td><center>53.19</center></td><td><center>43.87</center></td><td><center>51.59</center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-2024-06-28</td><td><center>50.56</center></td><td><center>44.70</center></td><td><center>52.19</center></td><td><center><strong>67.23</strong></center></td><td><center>57.69</center></td><td><center>30.23</center></td><td><center>51.34</center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-2024-10-09</td><td><center>52.21</center></td><td><center>47.94</center></td><td><center>53.50</center></td><td><center>66.06</center></td><td><center>59.72</center></td><td><center>40.16</center></td><td><center>45.90</center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-2025-04-23</td><td><center>54.66</center></td><td><center>50.31</center></td><td><center><strong>55.91</strong></center></td><td><center>67.01</center></td><td><center><strong>61.73</strong></center></td><td><center><strong>47.41</strong></center></td><td><center>45.61</center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-DPO-2024-10-09</td><td><center>49.96</center></td><td><center>46.29</center></td><td><center>53.29</center></td><td><center>65.57</center></td><td><center>58.15</center></td><td><center>34.77</center></td><td><center>41.70</center></td>
+</tr>
+<tr>
+<td><em>RoLlama3-8b-Instruct-DPO-2025-04-23</em></td><td><center><em><strong>55.86</strong></em></center></td><td><center><em><strong>52.26</strong></em></center></td><td><center><em>55.35</em></center></td><td><center><em>66.62</em></center></td><td><center><em>59.93</em></center></td><td><center><em>43.95</em></center></td><td><center><em><strong>57.06</strong></em></center></td>
+</tr>
+</tbody>
+</table>
+
+## Downstream tasks
+                           
+<table>
+<tbody>
+<tr>
+<td></td>
+<td colspan="4"><center><strong>LaRoSeDa</strong></center></td>
+<td colspan="4"><center><strong>WMT</strong></center></td>
+</tr>
+<tr>
+<td></td>
+<td colspan="2"><center><strong>Few-shot</strong></center></td>
+<td colspan="2"><center><strong>Finetuned</strong></center></td>
+<td colspan="2"><center><strong>Few-shot</strong></center></td>
+<td colspan="2"><center><strong>Finetuned</strong></center></td>
+</tr>
+<tr>
+<td><strong>Model</strong></td>
+<td><center><strong>Binary<br>(Macro F1)</strong></center></td>
+<td><center><strong>Multiclass<br>(Macro F1)</strong></center></td>
+<td><center><strong>Binary<br>(Macro F1)</strong></center></td>
+<td><center><strong>Multiclass<br>(Macro F1)</strong></center></td>
+<td><center><strong>EN-RO<br>(Bleu)</strong></center></td>
+<td><center><strong>RO-EN<br>(Bleu)</strong></center></td>
+<td><center><strong>EN-RO<br>(Bleu)</strong></center></td>
+<td><center><strong>RO-EN<br>(Bleu)</strong></center>
+</tr>
+<tr>
+<td>Llama-3-8B-Instruct</td><td><center>95.88</center></td><td><center>56.21</center></td><td><center><strong>98.53</strong></center></td><td><center>86.19</center></td><td><center>18.88</center></td><td><center><strong>30.98</strong></center></td><td><center><strong>28.02</strong></center></td><td><center>40.28</center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-2024-06-28</td><td><center>97.52</center></td><td><center><strong>67.41</strong></center></td><td><center>94.15</center></td><td><center>87.13</center></td><td><center><strong>24.01</strong></center></td><td><center>27.36</center></td><td><center>26.53</center></td><td><center>40.36</center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-2024-10-09</td><td><center>95.58</center></td><td><center>61.20</center></td><td><center>96.46</center></td><td><center><strong>87.26</strong></center></td><td><center>22.92</center></td><td><center>24.28</center></td><td><center>27.31</center></td><td><center><strong>40.52</strong></center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-2025-04-23</td><td><center>96.21</center></td><td><center>59.15</center></td><td><center>-</center></td><td><center>-</center></td><td><center>23.32</center></td><td><center>22.50</center></td><td><center>-</center></td><td><center>-</center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-DPO-2024-10-09</td><td><center>97.48</center></td><td><center>54.00</center></td><td><center>-</center></td><td><center>-</center></td><td><center>22.09</center></td><td><center>23.00</center></td><td><center>-</center></td><td><center>-</center></td>
+</tr>
+<tr>
+<td><em>RoLlama3-8b-Instruct-DPO-2025-04-23</em></td><td><center><em><strong>97.60</strong></em></center></td><td><center><em>62.16</em></center></td><td><center><em>-</em></center></td><td><center><em>-</em></center></td><td><center><em>18.14</em></center></td><td><center><em>14.13</em></center></td><td><center><em>-</em></center></td><td><center><em>-</em></center></td>
+</tr>
+</tbody>
+</table>
+
+
+<table>
+<tbody>
+<tr>
+<td></td>
+<td colspan="4"><center><strong>XQuAD</strong></center></td>
+<td colspan="4"><center><strong>STS</strong></center></td>
+</tr>
+<tr>
+<td></td>
+<td colspan="2"><center><strong>Few-shot</strong></center></td>
+<td colspan="2"><center><strong>Finetuned</strong></center></td>
+<td colspan="2"><center><strong>Few-shot</strong></center></td>
+<td colspan="2"><center><strong>Finetuned</strong></center></td>
+</tr>
+<tr>
+<td><strong>Model</strong></td>
+<td><center><strong>(EM)</strong></center></td>
+<td><center><strong>(F1)</strong></center></td>
+<td><center><strong>(EM)</strong></center></td>
+<td><center><strong>(F1)</strong></center></td>
+<td><center><strong>(Spearman)</strong></center></td>
+<td><center><strong>(Pearson)</strong></center></td>
+<td><center><strong>(Spearman)</strong></center></td>
+<td><center><strong>(Pearson)</strong></center></td>
+</tr>
+<tr>
+<td>Llama-3-8B-Instruct</td><td><center><strong>39.47</strong></center></td><td><center>58.67</center></td><td><center><strong>67.65</strong></center></td><td><center><strong>82.77</strong></center></td><td><center>73.04</center></td><td><center>72.36</center></td><td><center>83.49</center></td><td><center>84.06</center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-2024-06-28</td><td><center>39.43</center></td><td><center><strong>59.50</strong></center></td><td><center>44.45</center></td><td><center>59.76</center></td><td><center>77.20</center></td><td><center>77.87</center></td><td><center>85.80</center></td><td><center>86.05</center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-2024-10-09</td><td><center>18.89</center></td><td><center>31.79</center></td><td><center>50.84</center></td><td><center>65.18</center></td><td><center>77.60</center></td><td><center>76.86</center></td><td><center><strong>86.70</strong></center></td><td><center><strong>87.09</strong></center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-2025-04-23</td><td><center>11.01</center></td><td><center>23.55</center></td><td><center>-</center></td><td><center>-</center></td><td><center>76.78</center></td><td><center>74.36</center></td><td><center>-</center></td><td><center>-</center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-DPO-2024-10-09</td><td><center>26.05</center></td><td><center>42.77</center></td><td><center>-</center></td><td><center>-</center></td><td><center><strong>79.64</strong></center></td><td><center><strong>79.52</strong></center></td><td><center>-</center></td><td><center>-</center></td>
+</tr>
+<tr>
+<td><em>RoLlama3-8b-Instruct-DPO-2025-04-23</em></td><td><center><em>30.65</em></center></td><td><center><em>46.29</em></center></td><td><center><em>-</em></center></td><td><center><em>-</em></center></td><td><center><em>67.62</em></center></td><td><center><em>67.82</em></center></td><td><center><em>-</em></center></td><td><center><em>-</em></center></td>
+</tr>
+</tbody>
+</table>
+
+
+## Romanian MT-Bench
+
+<table>
+<tbody>
+<tr>
+<td><strong>Model</strong></td>
+<td><strong><center>Average</center></strong></td>
+<td><strong><center>1st turn</center></strong></td>
+<td><strong><center>2nd turn</center></strong></td>
+<td><strong><center>Answers in Ro</center></strong></td>
+</tr>
+<tr>
+<td>Llama-3-8B-Instruct</td><td><center>5.96</center></td><td><center>6.16</center></td><td><center>5.76</center></td><td><center>158/160</center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-2024-06-28</td><td><center>5.15</center></td><td><center>6.03</center></td><td><center>4.28</center></td><td><center><strong>160/160</strong></center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-2024-10-09</td><td><center>5.38</center></td><td><center>6.09</center></td><td><center>4.67</center></td><td><center><strong>160/160</strong></center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-2025-04-23</td><td><center>6.39</center></td><td><center><strong>7.12</strong></center></td><td><center>5.66</center></td><td><center><strong>160/160</strong></center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-DPO-2024-10-09</td><td><center>5.87</center></td><td><center>6.22</center></td><td><center>5.49</center></td><td><center><strong>160/160</strong></center></td>
+</tr>
+<tr>
+<td><em>RoLlama3-8b-Instruct-DPO-2025-04-23</em></td><td><center><em><strong>6.67</strong></em></center></td><td><center><em>6.81</em></center></td><td><center><em><strong>6.54</strong></em></center></td><td><center><em><strong>160/160</strong></em></center></td>
+</tr>
+</tbody>
+</table>
+
+
+## RoCulturaBench
+
+
+<table>
+<tbody>
+<tr>
+<td><strong>Model</strong></td>
+<td><strong><center>Average</center></strong></td>
+<td><strong><center>Answers in Ro</center></strong></td>
+</tr>
+<tr>
+<td>Llama-3-8B-Instruct</td><td><center>4.62</center></td><td><center><strong>100/100</strong></center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-2024-06-28</td><td><center>3.71</center></td><td><center><strong>100/100</strong></center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-2024-10-09</td><td><center>3.81</center></td><td><center><strong>100/100</strong></center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-2025-04-23</td><td><center>4.05</center></td><td><center><strong>100/100</strong></center></td>
+</tr>
+<tr>
+<td>RoLlama3-8b-Instruct-DPO-2024-10-09</td><td><center>4.40</center></td><td><center><strong>100/100</strong></center></td>
+</tr>
+<tr>
+<td><em>RoLlama3-8b-Instruct-DPO-2025-04-23</em></td><td><center><em><strong>4.83</strong></em></center></td><td><center><em><strong>100/100</strong></em></center></td>
+</tr>
+</tbody>
+</table>
+
+
+
+## RoLlama3 Model Family
+
+| Model              | Link  |
+|--------------------|:--------:|
+|RoLlama3-8b-Base-2024-05-14 | [link](https://huggingface.co/OpenLLM-Ro/RoLlam32-8b-Base-2024-05-14)    |
+|RoLlama3-8b-Instruct-2024-05-14 | [link](https://huggingface.co/OpenLLM-Ro/RoLlama3-8b-Instruct-2024-05-14)    |
+|RoLlama3-8b-Instruct-2024-10-09| [link](https://huggingface.co/OpenLLM-Ro/RoLlama3-8b-Instruct-2024-10-09) |
+|RoLlama3-8b-Instruct-2025-04-23| [link](https://huggingface.co/OpenLLM-Ro/RoLlama3-8b-Instruct-2025-04-23) |
+|RoLlama3-8b-Instruct-DPO-2024-10-09| [link](https://huggingface.co/OpenLLM-Ro/RoLlama3-8b-Instruct-DPO-2024-10-09) |
+|*RoLlama3-8b-Instruct-DPO-2025-04-23*| [link](https://huggingface.co/OpenLLM-Ro/RoLlama3-8b-Instruct-DPO-2025-04-23) |
+
+
+
+## Citation 
+
+```
+@misc{masala2024vorbecstiromanecsterecipetrain,
+      title={"Vorbe\c{s}ti Rom\^ane\c{s}te?" A Recipe to Train Powerful Romanian LLMs with English Instructions}, 
+      author={Mihai Masala and Denis C. Ilie-Ablachim and Alexandru Dima and Dragos Corlatescu and Miruna Zavelca and Ovio Olaru and Simina Terian-Dan and Andrei Terian-Dan and Marius Leordeanu and Horia Velicu and Marius Popescu and Mihai Dascalu and Traian Rebedea},
+      year={2024},
+      eprint={2406.18266},
+      archivePrefix={arXiv},
+      primaryClass={cs.CL},
+      url={https://arxiv.org/abs/2406.18266}, 
+}
+```
+<!-- **APA:**
+
 [More Information Needed]  -->