Instructions to use maicomputer/alpaca-native with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use maicomputer/alpaca-native with Transformers:

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

pipe = pipeline("text-generation", model="maicomputer/alpaca-native")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("maicomputer/alpaca-native")
model = AutoModelForCausalLM.from_pretrained("maicomputer/alpaca-native", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use maicomputer/alpaca-native with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/maicomputer/alpaca-native

SGLang

How to use maicomputer/alpaca-native 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 "maicomputer/alpaca-native" \
    --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": "maicomputer/alpaca-native",
		"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 "maicomputer/alpaca-native" \
        --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": "maicomputer/alpaca-native",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use maicomputer/alpaca-native with Docker Model Runner:
```
docker model run hf.co/maicomputer/alpaca-native
```

alpaca-native / README.md

chavinlo

Update README.md

cc7773c over 3 years ago

preview code

Raw History Blame

1.48 kB

Stanford Alpaca

This is a replica of Alpaca by Stanford' tatsu

Trained using the original instructions with a minor modification in FSDP mode

Other versions:

13B: https://huggingface.co/chavinlo/alpaca-13b

13B -> GPT4 : https://huggingface.co/chavinlo/gpt4-x-alpaca

Compute Used

Trained on 4xA100s for 6H Donated by redmond.ai

NO LORA HAS BEEN USED, this is a natively-finetuned model, hence "alpaca-native"

If you are interested on more llama-based models, you can check out my profile or search for other models at https://huggingface.co/models?other=llama

This (MIGHT) be a quantized version of this model, but be careful: https://boards.4channel.org/g/thread/92173062#p92182396

CONFIGURATION (default except fsdp):

torchrun --nproc_per_node=4 --master_port=3045 train.py \
    --model_name_or_path /workspace/llama-7b-hf \
    --data_path ./alpaca_data.json \
    --bf16 True \
    --output_dir /workspace/output \
    --num_train_epochs 3 \
    --per_device_train_batch_size 4 \
    --per_device_eval_batch_size 4 \
    --gradient_accumulation_steps 8 \
    --evaluation_strategy "no" \
    --save_strategy "steps" \
    --save_steps 200 \
    --save_total_limit 1 \
    --learning_rate 2e-5 \
    --weight_decay 0. \
    --warmup_ratio 0.03 \
    --lr_scheduler_type "cosine" \
    --logging_steps 1 \
    --fsdp "shard_grad_op auto_wrap" \
    --fsdp_transformer_layer_cls_to_wrap 'LLaMADecoderLayer' \
    --tf32 True --report_to="wandb"