Instructions to use Grogros/phi2-Instruct-reg1-1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Grogros/phi2-Instruct-reg1-1 with Transformers:

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

pipe = pipeline("text-generation", model="Grogros/phi2-Instruct-reg1-1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Grogros/phi2-Instruct-reg1-1")
model = AutoModelForCausalLM.from_pretrained("Grogros/phi2-Instruct-reg1-1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Grogros/phi2-Instruct-reg1-1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Grogros/phi2-Instruct-reg1-1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Grogros/phi2-Instruct-reg1-1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Grogros/phi2-Instruct-reg1-1

SGLang

How to use Grogros/phi2-Instruct-reg1-1 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 "Grogros/phi2-Instruct-reg1-1" \
    --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": "Grogros/phi2-Instruct-reg1-1",
		"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 "Grogros/phi2-Instruct-reg1-1" \
        --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": "Grogros/phi2-Instruct-reg1-1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Grogros/phi2-Instruct-reg1-1 with Docker Model Runner:
```
docker model run hf.co/Grogros/phi2-Instruct-reg1-1
```

phi2-Instruct-reg1-1 / finetuning_config.yaml

Grogros

Upload finetuning_config.yaml with huggingface_hub

592e0b4 verified 7 months ago

raw

history blame contribute delete

2.08 kB

	attn_implementation: sdpa
	backdoor_dataset: !!python/object/apply:src.data.dataset.DatasetType
	- AlpacaRefuseSmooth
	backdoor_dataset_mix_params: null
	balance_safecoder: false
	base_model: microsoft/phi-2
	dtype: bfloat16
	lora_config: null
	main_device: cuda:0
	meta_learning_configs:
	- dataset: !!python/object/apply:src.data.dataset.DatasetType
	- AlpacaGPT4
	device: cuda:2
	gradient_accumulation_steps: 1
	learning_rate: 5.0e-05
	lora_alpha: 32
	lora_r: 8
	loss_type: ce
	num_steps: 50
	optimizers:
	- adam
	per_device_batch_size: 1
	reg: 0.7
	run_every_n_steps: 1
	safecoder_lambda: 1.0
	sequence_length: 512
	use_lora: false
	warmup_steps: 0
	meta_learning_name: alpaca
	no_backdoor: false
	pgd_training_config: null
	precompute_distillation: false
	random_training_config:
	as_regularizer: false
	device: cuda:3
	loss_type: ce
	n_samples: 1
	norm: 5.0
	reg: 0.1
	safecoder_lambda: 1.0
	warmup_steps: 0
	reg_dataset: !!python/object/apply:src.data.dataset.DatasetType
	- SecretSauce
	reg_dataset_mix_params:
	? !!python/object/apply:src.data.dataset.DatasetType
	- AlpacaGPT4
	: 0.4
	? !!python/object/apply:src.data.dataset.DatasetType
	- AlpacaRefuseSmooth
	: 0.2
	? !!python/object/apply:src.data.dataset.DatasetType
	- OpenCoder
	: 0.2
	? !!python/object/apply:src.data.dataset.DatasetType
	- OpenMathInstruct
	: 0.2
	reg_device: cuda:1
	reg_lambda: 1.0
	reg_loss: distillation
	reg_model: eth-sri/phi-2-OurInstruct
	return_sublosses: false
	safecoder_lambda: 1.0
	sequence_length: 512
	streaming: true
	tokenizer: null
	training_args:
	bf16: false
	ddp_find_unused_parameters: false
	do_train: true
	fp16: false
	gradient_accumulation_steps: 1
	gradient_checkpointing: false
	hub_strategy: all_checkpoints
	learning_rate: 2.0e-05
	logging_steps: 10
	lr_scheduler_type: cosine
	max_steps: 2000
	num_train_epochs: 1
	optim: adafactor
	output_dir: Grogros/phi2-Instruct-reg1-1
	overwrite_output_dir: true
	per_device_train_batch_size: 32
	push_to_hub: true
	report_to: none
	save_steps: 2000
	save_strategy: steps
	warmup_ratio: 0.1