Instructions to use LarkAI/codet5p-770m_nl2sql_oig with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use LarkAI/codet5p-770m_nl2sql_oig with Transformers:

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

pipe = pipeline("text-generation", model="LarkAI/codet5p-770m_nl2sql_oig")

# Load model directly
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("LarkAI/codet5p-770m_nl2sql_oig")
model = AutoModelForSeq2SeqLM.from_pretrained("LarkAI/codet5p-770m_nl2sql_oig", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use LarkAI/codet5p-770m_nl2sql_oig with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "LarkAI/codet5p-770m_nl2sql_oig"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "LarkAI/codet5p-770m_nl2sql_oig",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/LarkAI/codet5p-770m_nl2sql_oig

SGLang

How to use LarkAI/codet5p-770m_nl2sql_oig 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 "LarkAI/codet5p-770m_nl2sql_oig" \
    --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": "LarkAI/codet5p-770m_nl2sql_oig",
		"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 "LarkAI/codet5p-770m_nl2sql_oig" \
        --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": "LarkAI/codet5p-770m_nl2sql_oig",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use LarkAI/codet5p-770m_nl2sql_oig with Docker Model Runner:
```
docker model run hf.co/LarkAI/codet5p-770m_nl2sql_oig
```

codet5p-770m_nl2sql_oig / README.md

JacksonLark

Update README.md

cb5e623 about 3 years ago

preview code

Raw

History Blame Contribute Delete

2.5 kB

	---
	license: apache-2.0
	datasets:
	- laion/OIG
	language:
	- en
	pipeline_tag: text2text-generation
	tags:
	- nl2sql
	widget:
	- text: 'Given the following schema:\ntrack (Track_ID, Name, Location, Seating, Year_Opened)\nrace (Race_ID, Name, Class, Date, Track_ID)\nWrite a SQL query to count the number of tracks.'
	example_title: 'count'
	- text: 'Given the following schema:\nmountain (Mountain_ID, Name, Height, Prominence, Range, Country)\nclimber (Climber_ID, Name, Country, Time, Points, Mountain_ID)\nWrite a SQL query to list the countries that have more than one mountain.'
	example_title: 'having'
	- text: 'Given the following schema:\nairports (apid, name, city, country, x, y, elevation, iata, icao)\nroutes (rid, dst_apid, dst_ap, src_apid, src_ap, alid, airline, codeshare)\nairlines (alid, name, iata, icao, callsign, country, active)\nWrite a SQL query to find the number of routes for each source airport and the airport name.'
	example_title: 'join'
	---

	# How to Use

	```python
	import torch
	from transformers import T5ForConditionalGeneration, AutoTokenizer

	device = torch.device("cuda:0")

	tokenizer = AutoTokenizer.from_pretrained("LarkAI/codet5p-770m_nl2sql_oig")
	model = T5ForConditionalGeneration.from_pretrained("LarkAI/codet5p-770m_nl2sql_oig").to(device)

	text = "Given the following schema:\ntrack (Track_ID, Name, Location, Seating, Year_Opened)\nrace (Race_ID, Name, Class, Date, Track_ID)\nWrite a SQL query to count the number of tracks."
	inputs = tokenizer.encode(text, return_tensors="pt").to(device)
	output_ids = model.generate(inputs, max_length=512)
	response_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
	# SELECT COUNT( * ) FROM track
	```

	# How to Train

	Dataset:
	- https://huggingface.co/datasets/laion/OIG#unified_sqlv1jsonl-17000
	- https://huggingface.co/datasets/laion/OIG#unified_sqlv2jsonl24000

	```json
	{
	"text":"<human>: Given the following schema:\nlocation (restaurant_id, house_number, street_name, city_name)\nrestaurant (id, name, food_type, city_name, rating)\ngeographic (city_name, county, region)\nWrite a SQL query to give me some good arabic -s on buchanan in san francisco ?\n<bot>: SELECT location.house_number , restaurant.name FROM location , restaurant WHERE location.city_name = \"san francisco\" AND location.street_name = \"buchanan\" AND restaurant.food_type = \"arabic\" AND restaurant.id = location.restaurant_id AND restaurant.rating > 2.5 ;",
	"metadata":{
	"source":"unified_sqlv1"
	}
	}
	```