Delete README.md

README.md

@@ -1,125 +0,0 @@
----
-base_model: silma-ai/SILMA-9B-Instruct-v1.0
-datasets:
-- MohammedNasser/ARabic_Reasoning_QA
-language:
-- ar
-library_name: transformers
-license: apache-2.0
-metrics:
-- accuracy
-pipeline_tag: question-answering
----
-
-
-
-
-# MohammedNasser/silma_9b_instruct_ft
-
-
-## Model Description
-
-The **silma_9b_instruct_ft** is a state-of-the-art language model fine-tuned specifically for Arabic reasoning tasks. This model excels in understanding and processing complex reasoning questions in Arabic, making it suitable for applications that require nuanced comprehension and logical inference. Leveraging advanced transformer architecture and extensive training, this model is designed to handle various reasoning challenges with high accuracy and efficiency.
-
-### Key Features
-- **Language**: Arabic
-- **Primary Task**: Reasoning
-- **Architecture**: Transformer-based, fine-tuned for Arabic language
-- **Sequence Length**: 20
-
-## Dataset
-
-This model was trained on the ARabic_Reasoning_QA dataset.
-
-- **Dataset Repository:** [ARabic_Reasoning_QA](https://huggingface.co/datasets/MohammedNasser/ARabic_Reasoning_QA)
-- **Dataset Description:** For detailed information about the dataset, please refer to the [README.md of ARabic_Reasoning_QA](https://huggingface.co/datasets/MohammedNasser/ARabic_Reasoning_QA/README.md).
-
-- 
-## Intended Uses & Limitations
-
-### Intended Uses
-- **Educational Tools**: Assist in creating intelligent tutoring systems and educational applications that require reasoning in Arabic.
-- **Research**: Facilitate research in Arabic natural language processing, especially in reasoning and inference tasks.
-- **Question Answering Systems**: Improve the accuracy of Arabic-based question-answering systems in various domains.
-
-### Limitations
-- **Training Data Scope**: Performance is dependent on the diversity and quality of the training data. May not generalize well to highly specialized domains or uncommon dialects.
-- **Single Epoch**: Trained for only one epoch; performance may improve with additional training.
-
-## Training and Evaluation Data
-
-### Training Data
-- **Dataset**: The model was fine-tuned using the ARabic Reasoning QA dataset.
-- **Content**: The dataset comprises 1000 reasoning questions across various difficulty levels, designed to test logical reasoning in Arabic.
-- **Source**: Custom dataset created for reasoning tasks, ensuring diverse and representative examples.
-
-### Evaluation Data
-- **Datasets Used**: Evaluated on `train`, `eval`, and `test` subsets of the ARabic Reasoning QA dataset.
-- **Metrics**: Accuracy and loss were measured to assess performance.
-- **Performance**: Achieved an evaluation loss of 0.038 
-
-## Training Procedure
-
-### Preprocessing
-- **Tokenization**: Text was tokenized using a pre-trained Arabic tokenizer.
-- **Sequence Length**: Text sequences were truncated or padded to a maximum length of 20 tokens.
-
-### Training
-- **Trainer**: Fine-tuned using the `SFTTrainer` class with the following parameters:
-  - **Gradient Accumulation**: Steps of 2
-  - **Learning Rate**: 2e-5
-  - **Optimizer**: AdamW with 32-bit precision
-  - **Gradient Clipping**: Max norm of 0.3
-  - **Warmup Ratio**: 0.03
-  - **Logging and Saving**: Logs and model checkpoints saved every 10 steps
-
-### Evaluation
-- **Evaluation Strategy**: Evaluated every 10 steps to monitor model performance on validation and test datasets.
-- **Metrics**: Accuracy and loss were recorded to assess the model's reasoning capabilities.
-
----
-
-
-
-### Training results
-
-| Training Loss | Epoch | Step | Validation Loss |
-|:-------------:|:-----:|:----:|:---------------:|
-| 2.1356        | 0.04  | 10   | 1.4071          |
-| 0.8079        | 0.08  | 20   | 0.2825          |
-| 0.1592        | 0.12  | 30   | 0.1427          |
-| 0.1202        | 0.16  | 40   | 0.1121          |
-| 0.1095        | 0.2   | 50   | 0.1071          |
-| 0.1024        | 0.24  | 60   | 0.1036          |
-| 0.0993        | 0.28  | 70   | 0.1002          |
-| 0.091         | 0.32  | 80   | 0.0992          |
-| 0.1096        | 0.36  | 90   | 0.0965          |
-| 0.0943        | 0.4   | 100  | 0.0916          |
-| 0.0882        | 0.44  | 110  | 0.0896          |
-| 0.0853        | 0.48  | 120  | 0.0848          |
-| 0.0767        | 0.52  | 130  | 0.0808          |
-| 0.0778        | 0.56  | 140  | 0.0765          |
-| 0.0698        | 0.6   | 150  | 0.0734          |
-| 0.0784        | 0.64  | 160  | 0.0694          |
-| 0.0648        | 0.68  | 170  | 0.0658          |
-| 0.0797        | 0.72  | 180  | 0.0630          |
-| 0.0591        | 0.76  | 190  | 0.0604          |
-| 0.0557        | 0.8   | 200  | 0.0582          |
-| 0.0567        | 0.84  | 210  | 0.0561          |
-| 0.057         | 0.88  | 220  | 0.0534          |
-| 0.0505        | 0.92  | 230  | 0.0515          |
-| 0.0483        | 0.96  | 240  | 0.0482          |
-| 0.0463        | 1.0   | 250  | 0.0463          |
-
-
-### Training Metrics
-[Training Loss on wandb 🔗](https://wandb.ai/mohnasgbr/huggingface/reports/train-loss-24-09-07-03-41-58---Vmlldzo5MjgxMTY4)
-
-
-### Framework versions
-
-- PEFT 0.12.1.dev0
-- Transformers 4.44.2
-- Pytorch 2.4.0+cu121
-- Datasets 2.21.0
-- Tokenizers 0.19.1