Instructions to use thedeba/gemma_bn_instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use thedeba/gemma_bn_instruct with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("INSAIT-Institute/BgGPT-Gemma-2-2.6B-IT-v1.0")
model = PeftModel.from_pretrained(base_model, "thedeba/gemma_bn_instruct")

Transformers

How to use thedeba/gemma_bn_instruct with Transformers:

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

pipe = pipeline("text-generation", model="thedeba/gemma_bn_instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("thedeba/gemma_bn_instruct", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use thedeba/gemma_bn_instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "thedeba/gemma_bn_instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "thedeba/gemma_bn_instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/thedeba/gemma_bn_instruct

SGLang

How to use thedeba/gemma_bn_instruct 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 "thedeba/gemma_bn_instruct" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "thedeba/gemma_bn_instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "thedeba/gemma_bn_instruct" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "thedeba/gemma_bn_instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use thedeba/gemma_bn_instruct with Docker Model Runner:
```
docker model run hf.co/thedeba/gemma_bn_instruct
```

gemma_bn_instruct / README.md

thedeba

End of training

b3ab80c verified 9 months ago

preview code

raw

history blame contribute delete

2.12 kB

	---
	library_name: peft
	license: gemma
	base_model: INSAIT-Institute/BgGPT-Gemma-2-2.6B-IT-v1.0
	tags:
	- base_model:adapter:INSAIT-Institute/BgGPT-Gemma-2-2.6B-IT-v1.0
	- lora
	- transformers
	pipeline_tag: text-generation
	model-index:
	- name: gemma_bn_instruct
	results: []
	---

	<!-- This model card has been generated automatically according to the information the Trainer had access to. You
	should probably proofread and complete it, then remove this comment. -->

	# gemma_bn_instruct

	This model is a fine-tuned version of [INSAIT-Institute/BgGPT-Gemma-2-2.6B-IT-v1.0](https://huggingface.co/INSAIT-Institute/BgGPT-Gemma-2-2.6B-IT-v1.0) on an unknown dataset.
	It achieves the following results on the evaluation set:
	- Loss: 1.7552

	## Model description

	More information needed

	## Intended uses & limitations

	More information needed

	## Training and evaluation data

	More information needed

	## Training procedure

	### Training hyperparameters

	The following hyperparameters were used during training:
	- learning_rate: 0.0002
	- train_batch_size: 8
	- eval_batch_size: 8
	- seed: 42
	- gradient_accumulation_steps: 8
	- total_train_batch_size: 64
	- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
	- lr_scheduler_type: linear
	- num_epochs: 3
	- mixed_precision_training: Native AMP

	### Training results

	\| Training Loss \| Epoch \| Step \| Validation Loss \|
	\|:-------------:\|:------:\|:----:\|:---------------:\|
	\| 2.0629 \| 0.2844 \| 200 \| 1.9151 \|
	\| 1.8976 \| 0.5689 \| 400 \| 1.8619 \|
	\| 1.8624 \| 0.8533 \| 600 \| 1.8252 \|
	\| 1.787 \| 1.1365 \| 800 \| 1.8043 \|
	\| 1.7492 \| 1.4210 \| 1000 \| 1.7880 \|
	\| 1.7227 \| 1.7054 \| 1200 \| 1.7763 \|
	\| 1.7327 \| 1.9899 \| 1400 \| 1.7640 \|
	\| 1.6352 \| 2.2731 \| 1600 \| 1.7663 \|
	\| 1.6366 \| 2.5575 \| 1800 \| 1.7605 \|
	\| 1.6381 \| 2.8420 \| 2000 \| 1.7552 \|


	### Framework versions

	- PEFT 0.17.0
	- Transformers 4.51.3
	- Pytorch 2.7.0+cu126
	- Datasets 4.0.0
	- Tokenizers 0.21.1