Update README.md

README.md

@@ -1,16 +1,16 @@
 ---
 license: apache-2.0
 datasets:
-- avemio/German_RAG-CPT-HESSIAN-AI
-- avemio/German_RAG-SFT-ShareGPT-HESSIAN-AI
-- avemio/German_RAG-ORPO-ShareGPT-HESSIAN-AI
+- avemio/German-RAG-CPT-HESSIAN-AI
+- avemio/German-RAG-SFT-ShareGPT-HESSIAN-AI
+- avemio/German-RAG-ORPO-ShareGPT-HESSIAN-AI
 - VAGOsolutions/SauerkrautLM-Fermented-GER-DPO
 - VAGOsolutions/SauerkrautLM-Fermented-Irrelevance-GER-DPO
 language:
 - en
 - de
 base_model:
-- avemio/German_RAG-NEMO-12B-SFT-HESSIAN-AI
+- avemio/German-RAG-NEMO-12B-SFT-HESSIAN-AI
 pipeline_tag: question-answering
 tags:
 - German
@@ -22,25 +22,25 @@ tags:
 ---
 
 
-<img src="https://www.German_RAG.ai/wp-content/uploads/2024/12/German_RAG-ICON-TO-WORDLOGO-Animation_Loop-small-ezgif.com-video-to-gif-converter.gif" alt="German_RAG Logo" width="400" style="margin-left:'auto' margin-right:'auto' display:'block'"/>
+<img src="https://www.German-RAG.ai/wp-content/uploads/2024/12/German-RAG-ICON-TO-WORDLOGO-Animation_Loop-small-ezgif.com-video-to-gif-converter.gif" alt="German-RAG Logo" width="400" style="margin-left:'auto' margin-right:'auto' display:'block'"/>
 
 
-# German_RAG-NEMO-12B-ORPO-HESSIAN-AI
+# German-RAG-NEMO-12B-ORPO-HESSIAN-AI
 
 <!-- Provide a quick summary of what the model is/does. -->
 
-**German_RAG** (**G**erman **R**etrieval **A**ugmented **G**eneration) models are designed for the German-speaking market, enabling innovation and AI solutions to drive German research collaboration in business-focused Generative AI by 2025
+**German-RAG** (**G**erman **R**etrieval **A**ugmented **G**eneration) models are designed for the German-speaking market, enabling innovation and AI solutions to drive German research collaboration in business-focused Generative AI by 2025
 
-Our German_RAG-NEMP-ORPO model are trained on this **[German_RAG-ORPO](https://huggingface.co/datasets/avemio/German_RAG-ORPO-ShareGPT-HESSIAN-AI) dataset.**
+Our German-RAG-NEMP-ORPO model are trained on this **[German-RAG-ORPO](https://huggingface.co/datasets/avemio/German-RAG-ORPO-ShareGPT-HESSIAN-AI) dataset.**
 
 ## Model Details
 
 The core models released in this batch are the following: 
 | Size | Training Tokens | 
 |------|--------|
-| [German_RAG-NEMO-CPT](https://huggingface.co/avemio/German_RAG-NEMO-12B-CPT-HESSIAN-AI)   | 507.47 million |
-| [German_RAG-NEMO-SFT](https://huggingface.co/avemio/German_RAG-NEMO-12B-SFT-HESSIAN-AI) |  2.03 billion  |  
-| [German_RAG-NEMO-ORPO](https://huggingface.co/avemio/German_RAG-NEMO-12B-ORPO-HESSIAN-AI) |  2.0577 billion  | 
+| [German-RAG-NEMO-CPT](https://huggingface.co/avemio/German-RAG-NEMO-12B-CPT-HESSIAN-AI)   | 507.47 million |
+| [German-RAG-NEMO-SFT](https://huggingface.co/avemio/German-RAG-NEMO-12B-SFT-HESSIAN-AI) |  2.03 billion  |  
+| [German-RAG-NEMO-ORPO](https://huggingface.co/avemio/German-RAG-NEMO-12B-ORPO-HESSIAN-AI) |  2.0577 billion  | 
 ### Model Description
 
 <!-- Provide a longer summary of what this model is. -->
@@ -50,19 +50,19 @@ The core models released in this batch are the following:
 - **Model type:** a Transformer style autoregressive language model.
 - **Language(s) (NLP):** German, English
 - **License:** The code and model are released under Apache 2.0.
-- **Contact:** [German_RAG@avemio.digital](mailto:German_RAG@avemio.digital)
+- **Contact:** [German-RAG@avemio.digital](mailto:German-RAG@avemio.digital)
 
 
 ### Model Sources
 
 <!-- Provide the basic links for the model. -->
 
-- **Training Study:** [Training Study](https://avemio.digital/wp-content/uploads/2025/01/German_RAG-TRAINING-STUDY-Advancing-German-Language-AI-with-hessian-AI.pdf)
+- **Training Study:** [Training Study](https://avemio.digital/wp-content/uploads/2025/01/German-RAG-TRAINING-STUDY-Advancing-German-Language-AI-with-hessian-AI.pdf)
 - **Repositories:** 
     - Training: [Colab-Notebook](https://colab.research.google.com/drive/18SH_aYLCnw1K7cRGOTTZ80y98V5Kquxb?usp=sharing)
     - Evaluation code: 
-        - [German_RAG-LLM-HARD-BENCHMARK](https://github.com/avemio-digital/German_RAG-LLM-HARD-BENCHMARK.git)
-        - [German_RAG-LLM-EASY-BENCHMARK](https://github.com/avemio-digital/German_RAG-LLM-EASY-BENCHMARK.git)
+        - [German-RAG-LLM-HARD-BENCHMARK](https://github.com/avemio-digital/German-RAG-LLM-HARD-BENCHMARK.git)
+        - [German-RAG-LLM-EASY-BENCHMARK](https://github.com/avemio-digital/German-RAG-LLM-EASY-BENCHMARK.git)
 -  **Technical blog post:**
 <!-- - **Press release:** TODO -->
 
@@ -76,7 +76,7 @@ Now, proceed as usual with HuggingFace:
 ```python
 from transformers import AutoModelForCausalLM, AutoTokenizer
  
-model_name = "avemio/German_RAG-NEMO-12B-ORPO-HESSIAN-AI"
+model_name = "avemio/German-RAG-NEMO-12B-ORPO-HESSIAN-AI"
  
 model = AutoModelForCausalLM.from_pretrained(
     model_name,
@@ -125,7 +125,7 @@ response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
 We are providing a comprehensive Google Colab notebook to guide users through the process of fine-tuning our model, complete with detailed instructions, essential dependencies, and configurable settings.
  [Colab-Notebook](https://colab.research.google.com/drive/18SH_aYLCnw1K7cRGOTTZ80y98V5Kquxb?usp=sharing).
 
-## German_RAG-LLM-EASY-BENCHMARK EVAL
+## German-RAG-LLM-EASY-BENCHMARK EVAL
 
 <!-- This section describes the evaluation protocols and provides the results. -->
 The evaluation was performed using seven subsets, focusing on extraction recall, question answering (QA) with multiple references, and time difference reasoning. Relevant context and summarization were treated as distinct subsets, each playing a crucial role in the evaluation process. For relevant context, the model's ability to identify and extract pertinent information from the source material was assessed. In contrast, the summarization subset evaluated the model's capability to generate concise and accurate summaries based on the relevant context.
@@ -138,7 +138,7 @@ Four evaluation metrics were employed across all subsets: language quality, over
 -   **Overall score:** This metric combined the results from the previous three metrics, offering a comprehensive evaluation of the model's capabilities across all subsets.
 
 
-| Metric                                    | [Vanila-Mistral-Nemo-Instruct-2407](https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407) | [German_RAG-NEMO-SFT](https://huggingface.co/avemio/German_RAG-NEMO-12B-SFT-HESSIAN-AI) | **[German_RAG-NEMO-ORPO](https://huggingface.co/avemio/German_RAG-NEMO-12B-ORPO-HESSIAN-AI)** | GPT-3.5-TURBO | 
+| Metric                                    | [Vanila-Mistral-Nemo-Instruct-2407](https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407) | [German-RAG-NEMO-SFT](https://huggingface.co/avemio/German-RAG-NEMO-12B-SFT-HESSIAN-AI) | **[German-RAG-NEMO-ORPO](https://huggingface.co/avemio/German-RAG-NEMO-12B-ORPO-HESSIAN-AI)** | GPT-3.5-TURBO | 
 |------------------------------------------|---------------------------------------------------------------------------------|--------------------------------------------------------------------------------|------------------------------------------------------------------------------------------------|----------------|
 | Average Language Quality             | 85.88                                                                          | 89.61                                                                     | **89.1**                                                                                          | 91.86                                            |
 | **OVERALL SCORES (weighted):**       |                                                                            |                                                                            |                                                                                           |                                            |
@@ -150,11 +150,11 @@ Four evaluation metrics were employed across all subsets: language quality, over
 | summarizations         | 73.8                                                                           | 81.6                                                                           | **80.3**                                                                                          |                     86.9                        |
 
 
-## German_RAG-LLM-HARD-BENCHMARK EVAL
+## German-RAG-LLM-HARD-BENCHMARK EVAL
 
-<img src="https://avemio.digital/wp-content/uploads/2025/01/German_RAG-NEMO-ORPO.png" alt="German_RAG Logo" width="600" style="margin-left:'auto' margin-right:'auto' display:'block'"/>
+<img src="https://avemio.digital/wp-content/uploads/2025/01/German-RAG-NEMO-ORPO.png" alt="German-RAG Logo" width="600" style="margin-left:'auto' margin-right:'auto' display:'block'"/>
 
-| Metric                  | [Vanila-Mistral-Nemo-Instruct-2407](https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407) | **[German_RAG-NEMO-ORPO](https://huggingface.co/avemio/German_RAG-NEMO-12B-ORPO-HESSIAN-AI)** | GPT-3.5-TURBO | GPT-4o | GPT-4o-mini |
+| Metric                  | [Vanila-Mistral-Nemo-Instruct-2407](https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407) | **[German-RAG-NEMO-ORPO](https://huggingface.co/avemio/German-RAG-NEMO-12B-ORPO-HESSIAN-AI)** | GPT-3.5-TURBO | GPT-4o | GPT-4o-mini |
 |-------------------------|-----------------------------------------------------------------------------------------|---------------------------------------------------------------------------------------|----------------|---------|-------------|
 | **OVERALL SCORES (weighted):** |                                                                                 |                                                                                       |                |         |             |
 | hard_reasoning_de       | 43.6                                                                                    | **49.7**                                                                                  | 37.9           | 62.9    | 58.4        |
@@ -164,7 +164,7 @@ Four evaluation metrics were employed across all subsets: language quality, over
 ## Model Details
 
 ### Data
-For training data details, please see the [German_RAG-ORPO-Dataset](https://huggingface.co/datasets/avemio/German_RAG-ORPO-ShareGPT-HESSIAN-AI) documentation.
+For training data details, please see the [German-RAG-ORPO-Dataset](https://huggingface.co/datasets/avemio/German-RAG-ORPO-ShareGPT-HESSIAN-AI) documentation.
 
 The ORPO Tasks Dataset represents a specialized collection for fine-tuning language models with a focus on RAG-specific capabilities. 
 
@@ -217,7 +217,7 @@ The subsets can be for this training step are derived from 3 different sources:
 ### Architecture
 
 
-| Parameter            | German_RAG-NEMO-ORPO                                                                                   |
+| Parameter            | German-RAG-NEMO-ORPO                                                                                   |
 |-----------------------|-----------------------------------------------------------------------------------------------|
 | **d_model**          | 5120                                                                                         |
 | **num heads**        | 32                                                                                            |
@@ -235,7 +235,7 @@ The subsets can be for this training step are derived from 3 different sources:
 ### Hyperparameters 
 
 
-| Parameter                 | German_RAG-NEMO-ORPO       |
+| Parameter                 | German-RAG-NEMO-ORPO       |
 |---------------------------|--------------------|
 | **warmup steps**          | 50                |
 | **peak LR**               | 5.0E-07           |
@@ -246,19 +246,19 @@ The subsets can be for this training step are derived from 3 different sources:
 
 ## Environmental Impact
 
-German_RAG-NEMO-ORPO, running on NVIDIA A100 with 80 GPUs for 4 days, has an approximate power consumption as follows:
+German-RAG-NEMO-ORPO, running on NVIDIA A100 with 80 GPUs for 4 days, has an approximate power consumption as follows:
 
 It's important to note that the actual power consumption may vary depending on the specific workload and operational conditions. For accurate power consumption measurements, using dedicated power monitoring tools is recommended.
 
 | Model          | GPU Type            | Power Consumption From GPUs | 
 |----------------|---------------------|-----------------------------|
-| German_RAG-NEMO-ORPO   | A100 ([Hessian AI supercomputer](https://hessian.ai/de/)) | 0.01843 MWh                    |
+| German-RAG-NEMO-ORPO   | A100 ([Hessian AI supercomputer](https://hessian.ai/de/)) | 0.01843 MWh                    |
 ## Bias, Risks, and Limitations
 
 Like any base language model or fine-tuned model without safety filtering, it is relatively easy for a user to prompt these models to generate harmful and generally sensitive content.
 Such content can also be produced unintentionally, especially in the case of bias, so we recommend users consider the risks of applications of this technology.
 
-Otherwise, many facts from German_RAG-NEMO-ORPO or any LLM will often not be true, so they should be checked.
+Otherwise, many facts from German-RAG-NEMO-ORPO or any LLM will often not be true, so they should be checked.
 
 
 
@@ -266,9 +266,9 @@ Otherwise, many facts from German_RAG-NEMO-ORPO or any LLM will often not be tru
 ## Model Card Contact
 
 
-For errors in this model card, please contact ([German_RAG@avemio.digital](mailto:German_RAG@avemio.digital)).
+For errors in this model card, please contact ([German-RAG@avemio.digital](mailto:German-RAG@avemio.digital)).
 
-## The German_RAG AI Team
+## The German-RAG AI Team
 [Marcel Rosiak](https://de.linkedin.com/in/marcel-rosiak)
 [Soumya Paul](https://de.linkedin.com/in/soumya-paul-1636a68a)
 [Siavash Mollaebrahim](https://de.linkedin.com/in/siavash-mollaebrahim-4084b5153?trk=people-guest_people_search-card)