Instructions to use Qybera/dakitari-instruct-v2-advanced with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Qybera/dakitari-instruct-v2-advanced with Adapters:

from adapters import AutoAdapterModel

model = AutoAdapterModel.from_pretrained("undefined")
model.load_adapter("Qybera/dakitari-instruct-v2-advanced", set_active=True)

Transformers

How to use Qybera/dakitari-instruct-v2-advanced with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Qybera/dakitari-instruct-v2-advanced")

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("Qybera/dakitari-instruct-v2-advanced", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Qybera/dakitari-instruct-v2-advanced with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Qybera/dakitari-instruct-v2-advanced"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Qybera/dakitari-instruct-v2-advanced",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Qybera/dakitari-instruct-v2-advanced

SGLang

How to use Qybera/dakitari-instruct-v2-advanced with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Qybera/dakitari-instruct-v2-advanced" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Qybera/dakitari-instruct-v2-advanced",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Qybera/dakitari-instruct-v2-advanced" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Qybera/dakitari-instruct-v2-advanced",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Qybera/dakitari-instruct-v2-advanced with Docker Model Runner:
```
docker model run hf.co/Qybera/dakitari-instruct-v2-advanced
```

YAML Metadata Warning:The pipeline tag "text2text-generation" is not in the official list: text-classification, token-classification, table-question-answering, question-answering, zero-shot-classification, translation, summarization, feature-extraction, text-generation, fill-mask, sentence-similarity, text-to-speech, text-to-audio, automatic-speech-recognition, audio-to-audio, audio-classification, audio-text-to-text, voice-activity-detection, depth-estimation, image-classification, object-detection, image-segmentation, text-to-image, image-to-text, image-to-image, image-to-video, unconditional-image-generation, video-classification, reinforcement-learning, robotics, tabular-classification, tabular-regression, tabular-to-text, table-to-text, multiple-choice, text-ranking, text-retrieval, time-series-forecasting, text-to-video, image-text-to-text, image-text-to-image, image-text-to-video, visual-question-answering, document-question-answering, zero-shot-image-classification, graph-ml, mask-generation, zero-shot-object-detection, text-to-3d, image-to-3d, image-feature-extraction, video-text-to-text, keypoint-detection, visual-document-retrieval, any-to-any, video-to-video, other

Dakitari-Instruct Medical Language Model

A specialized language model for medical text generation and question answering, trained on PubMed abstracts and medical QA datasets.

Model Description

Model Type: Transformer-based Language Model
Language: English
License: MIT
Training Data: PubMed abstracts + Medical QA pairs

Usage

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("path/to/dakitari-instruct")
model = AutoModelForCausalLM.from_pretrained("path/to/dakitari-instruct")

Training Data

This model is trained on:

PubMed abstracts (biomedical literature)
Medical question-answer pairs

For detailed dataset information, see dataset_card.md.

Features

Medical text generation
Medical question answering
Research assistance
Built on transformer architecture
Trained on PubMed abstracts and medical Q&A datasets

Requirements

Python 3.8+
TensorFlow 2.13+
Transformers library
Other dependencies listed in requirements.txt

Installation

Clone the repository:

git clone https://github.com/elijahnzeli1/dakitari-instruct.git
cd dakitari-instruct

Create a virtual environment (recommended):

python -m venv venv
#for windows
.\venv\Scripts\Activate
#for mac/linux
source venv/bin/activate

Install dependencies:

pip install -r requirements.txt

Training the Model

Start training:

python train.py --batch_size 32 --epochs 10

Monitor training progress with TensorBoard:

tensorboard --logdir logs

Then open your browser and navigate to http://localhost:6006 to view training metrics.

Training logs are also saved in CSV format in the logs directory for backup.

Training Parameters

You can customize the training by adjusting these parameters:

--batch_size: Batch size for training (default: 32)
--epochs: Number of training epochs (default: 10)
--max_length: Maximum sequence length (default: 512)
--embed_dim: Embedding dimension (default: 256)
--num_heads: Number of attention heads (default: 8)
--ff_dim: Feed-forward dimension (default: 512)
--num_transformer_blocks: Number of transformer blocks (default: 6)
--dropout_rate: Dropout rate (default: 0.1)
--learning_rate: Learning rate (default: 1e-4)
--checkpoint_dir: Directory to save model checkpoints (default: "checkpoints")
--log_dir: Directory to save training logs (default: "logs")

Project Structure

dakitari-instruct/
├── data/
│   └── preprocess.py      # Data preprocessing utilities
├── model/
│   └── transformer_model.py # Model architecture
├── train.py              # Training script
├── requirements.txt      # Project dependencies
└── README.md            # This file

Hardware Requirements

Minimum: 16GB RAM, NVIDIA GPU with 8GB VRAM
Recommended: 32GB RAM, NVIDIA GPU with 16GB+ VRAM

Dataset Sources

The model is trained on:

PubMed abstracts
Medical questions and answers pairs

Contributing

Fork the repository
Create your feature branch
Commit your changes
Push to the branch
Create a new Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

Thanks to the HuggingFace team for the transformers library
Thanks to the TensorFlow team for the excellent framework
Thanks to the medical community for the valuable datasets

Now you can track your training progress locally using TensorBoard, which provides:

Real-time metrics visualization
Learning rate tracking
Model graph visualization
Histogram of weights and biases

The training metrics are also saved in CSV format as a backup, which you can analyze using any spreadsheet software or data analysis tools. To view the training progress:

Start training your model
In a separate terminal, run tensorboard --logdir logs
Open http://localhost:6006 in your browser

Downloads last month: -

Safetensors

Model size

0.1B params

Tensor type

F32

Model tree for Qybera/dakitari-instruct-v2-advanced

Base model

Qybera/dakitari-instruct-v1.0

Adapter

(1)

this model

Qybera
/

dakitari-instruct-v2-advanced