Instructions to use markm39/openproof-tactic-2b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use markm39/openproof-tactic-2b with Transformers:

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

pipe = pipeline("text-generation", model="markm39/openproof-tactic-2b")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("markm39/openproof-tactic-2b", dtype="auto")

PEFT
How to use markm39/openproof-tactic-2b with PEFT:
```
Task type is invalid.
```
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use markm39/openproof-tactic-2b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "markm39/openproof-tactic-2b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "markm39/openproof-tactic-2b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/markm39/openproof-tactic-2b

SGLang

How to use markm39/openproof-tactic-2b 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 "markm39/openproof-tactic-2b" \
    --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": "markm39/openproof-tactic-2b",
		"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 "markm39/openproof-tactic-2b" \
        --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": "markm39/openproof-tactic-2b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use markm39/openproof-tactic-2b with Docker Model Runner:
```
docker model run hf.co/markm39/openproof-tactic-2b
```

openproof-tactic-2b / README.md

markm39

Update README.md

26b4c66 verified about 1 month ago

preview code

raw

history blame contribute delete

3.32 kB

	---
	base_model: Qwen/Qwen3.5-2B
	pipeline_tag: text-generation
	tags:
	- base_model:adapter:Qwen/Qwen3.5-2B
	- lora
	- transformers
	- lean4
	- theorem-proving
	- tactic-prediction
	- formal-verification
	- mathlib
	- peft
	license: mit
	datasets:
	- markm39/openproof-tactic-pairs
	---

	# OpenProof Tactic 2B

	A Lean 4 tactic prediction model fine-tuned from [Qwen/Qwen3.5-2B](https://huggingface.co/Qwen/Qwen3.5-2B) on 190K+ Mathlib tactic pairs.

	Given a Lean 4 proof goal state, the model predicts the next tactic to apply. Designed for use in best-first search (BFS) proof engines like [OpenProof](https://github.com/mxthematic/openproof).

	## Usage

	```python
	from transformers import AutoModelForCausalLM, AutoTokenizer
	from peft import PeftModel
	import torch

	base = "Qwen/Qwen3.5-2B"
	tokenizer = AutoTokenizer.from_pretrained(base)
	model = AutoModelForCausalLM.from_pretrained(base, torch_dtype=torch.bfloat16, device_map="cuda")
	model = PeftModel.from_pretrained(model, "markm39/openproof-tactic-2b")

	# Input: goal state followed by :::
	prompt = "n m : Nat\\n⊢ n + m = m + n:::"
	inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
	with torch.no_grad():
	out = model.generate(**inputs, max_new_tokens=64, do_sample=True, temperature=0.6, top_k=10)
	print(tokenizer.decode(out[0][inputs.input_ids.shape[1]:], skip_special_tokens=True))
	# Output: induction m with \| zero => simp \| succ m ih => simp [Nat.add_assoc, ih, Nat.add_comm]
	```

	## Training

	- Base model: Qwen/Qwen3.5-2B
	- Method: SFT with LoRA (rank 64, alpha 128)
	- Trainable params: 43.6M / 1.93B (2.3%)
	- Data: [markm39/openproof-tactic-pairs](https://huggingface.co/datasets/markm39/openproof-tactic-pairs) -- 190K+ Lean 4 goal-tactic pairs extracted from Mathlib
	- Input format: `{goal_state}:::`
	- Output format: tactic string
	- Epochs: ~1.8 (12,500 steps)
	- Batch size: effective 128 (2 per device x 32 gradient accumulation)
	- Learning rate: 2e-5 cosine schedule
	- Best eval loss: 0.5492 (checkpoint-12500)
	- Hardware: RunPod GPU instance
	- Framework: HuggingFace TRL + PEFT

	## Intended Use

	Tactic prediction for automated theorem proving in Lean 4. The model generates candidate tactics given a proof goal state, which are then verified by the Lean type checker. Designed to be used with beam search -- sample multiple candidates and verify each.

	Not intended for general-purpose text generation.

	## Limitations

	- Trained on Mathlib tactics only -- may not generalize to custom Lean libraries
	- LoRA adapter (must be loaded on top of Qwen3.5-2B base)
	- SFT only (no RL/GRPO refinement yet)
	- May hallucinate lemma names that don't exist in Mathlib

	## Links

	- [OpenProof](https://github.com/mxthematic/openproof) -- the theorem prover that uses this model
	- [Training code](https://github.com/mxthematic/openproof-ml)
	- [Training data](https://huggingface.co/datasets/markm39/openproof-tactic-pairs)

	## Acknowledgments

	Training data derived from [LeanDojo](https://leandojo.org/) (Yang et al.), [Lean Workbook](https://huggingface.co/datasets/internlm/Lean-Workbook) (InternLM), and [Goedel-Pset](https://huggingface.co/datasets/Goedel-LM/Goedel-Pset) (Goedel-LM). Tactic extraction from whole proofs uses [Pantograph](https://github.com/leanprover/Pantograph) kernel-level invocations.