Instructions to use thanhdath/FINER-SQL-0.5B-BIRD with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use thanhdath/FINER-SQL-0.5B-BIRD with Transformers:

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

pipe = pipeline("text-generation", model="thanhdath/FINER-SQL-0.5B-BIRD")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("thanhdath/FINER-SQL-0.5B-BIRD")
model = AutoModelForCausalLM.from_pretrained("thanhdath/FINER-SQL-0.5B-BIRD")
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

vLLM

How to use thanhdath/FINER-SQL-0.5B-BIRD with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "thanhdath/FINER-SQL-0.5B-BIRD"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "thanhdath/FINER-SQL-0.5B-BIRD",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/thanhdath/FINER-SQL-0.5B-BIRD

SGLang

How to use thanhdath/FINER-SQL-0.5B-BIRD 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 "thanhdath/FINER-SQL-0.5B-BIRD" \
    --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": "thanhdath/FINER-SQL-0.5B-BIRD",
		"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 "thanhdath/FINER-SQL-0.5B-BIRD" \
        --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": "thanhdath/FINER-SQL-0.5B-BIRD",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use thanhdath/FINER-SQL-0.5B-BIRD with Docker Model Runner:
```
docker model run hf.co/thanhdath/FINER-SQL-0.5B-BIRD
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

FINER-SQL-0.5B-BIRD

A small but capable 0.5 B-parameter Text-to-SQL model fine-tuned from griffith-bigdata/Qwen-2.5-Coder-0.5B-SQL-Writer with GRPO + the FINER-SQL dense rewards (Memory + Atomic).

✅ 50.85% Execution Accuracy on BIRD Dev (n=30, value-aware voting). Runs on a 4-8 GB GPU.

📄 See other models: https://huggingface.co/collections/griffith-bigdata/finer-sql 📄 GitHub: https://github.com/thanhdath/finer-sql/tree/main

FINER-SQL Model Family — Comparison Across All Sizes

Model	Params	BIRD Dev (n=30, vav)	Spider Dev (n=30, vav, +agg_hint)
FINER-SQL-3B-BIRD	3 B	67.54% ✅	83.8%
FINER-SQL-3B-Spider	3 B	63.04%	85.10% ✅
FINER-SQL-0.5B-BIRD (this model)	0.5 B	50.85% ✅	68.6%
FINER-SQL-0.5B-Spider	0.5 B	TBD	75.0% ✅

The 0.5 B family demonstrates that GRPO + FINER rewards scale down to deployment-friendly sizes while retaining most of the gain.

Inference

Quick start (vLLM)

from vllm import LLM, SamplingParams

llm = LLM(
    model="griffith-bigdata/FINER-SQL-0.5B-BIRD",
    dtype="bfloat16",
    max_model_len=4096,
    gpu_memory_utilization=0.7,
)

system_prompt = """You are a meticulous SQL expert. Generate a single, correct SQL query for the user question and the provided database schema.
Follow this exact response format:

Rules:
- Output exactly one SQL statement.
- The SQL must be executable on SQLite.
- Do not include any explanatory text.
- Output one SQL statement only. Do not include any extra text, tags, or code fences."""

sampling = SamplingParams(n=30, temperature=1.0, max_tokens=2048)
messages = [
    {"role": "system", "content": system_prompt},
    {"role": "user", "content": f"Database Schema:\n{schema}\n\nQuestion: {question}\n\nEvidence: {evidence}"},
]
output = llm.chat(messages, sampling)
candidate_sqls = [c.text.split("</think>")[-1].strip() for c in output[0].outputs]
# Apply majority voting (vav) — see GitHub repo

Recommended evaluation pipeline

Generate n=30 candidates with temperature=1.0
Execute each candidate; group results
Pick from the largest non-empty success group (value-aware voting, "vav")
Score with the official BIRD evaluator

This pipeline gives 50.85% MV on BIRD Dev V2 prompts (best 0.5 B result).

Detailed BIRD Dev results (V2 prompts, n=30, vav)

Difficulty	Count	Execution Accuracy
Simple	925	~58%
Moderate	464	~42%
Challenging	145	~38%
All	1534	50.85%

Recall@30: 68.32% (any-correct rate among 30 candidates).

Cross-benchmark: this model on Spider Dev (zero-shot)

Setup	Spider Official EX
Default	68.6%
FINER-SQL-0.5B-Spider (specialist)	75.0%

For Spider use-cases, the FINER-SQL-0.5B-Spider checkpoint is preferred (+6.4 pp).

Training

Parameter	Value
Base model	`griffith-bigdata/Qwen-2.5-Coder-0.5B-SQL-Writer`
Algorithm	GRPO
Train data	BIRD train (V2 prompts, top-30 GRAST)
Total steps	4000 (this checkpoint = 3000)
Learning rate	8e-6
Num generations per prompt	32
Gradient accumulation	32
Max completion length	2048
Max prompt length	2048
Temperature (rollout)	1.0
Selection during eval	vav (value-aware voting)
Rewards	Execution + Atomic + Memory + Format
Intrinsic Top-K	20 (ChromaDB)

License

Inherits the base model's license (Apache 2.0).

Citation

@article{finer-sql-2026,
  title  = {FINER-SQL: Fine-grained reasoning rewards for small Text-to-SQL models},
  author = {Thanh Dat and others},
  year   = {2026},
}