Instructions to use juwon1105/RLCR-phi4mini-bigmathdigits5000 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use juwon1105/RLCR-phi4mini-bigmathdigits5000 with Transformers:

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

pipe = pipeline("text-generation", model="juwon1105/RLCR-phi4mini-bigmathdigits5000", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

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

tokenizer = AutoTokenizer.from_pretrained("juwon1105/RLCR-phi4mini-bigmathdigits5000", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("juwon1105/RLCR-phi4mini-bigmathdigits5000", trust_remote_code=True, device_map="auto")
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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use juwon1105/RLCR-phi4mini-bigmathdigits5000 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "juwon1105/RLCR-phi4mini-bigmathdigits5000"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "juwon1105/RLCR-phi4mini-bigmathdigits5000",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/juwon1105/RLCR-phi4mini-bigmathdigits5000

SGLang

How to use juwon1105/RLCR-phi4mini-bigmathdigits5000 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 "juwon1105/RLCR-phi4mini-bigmathdigits5000" \
    --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": "juwon1105/RLCR-phi4mini-bigmathdigits5000",
		"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 "juwon1105/RLCR-phi4mini-bigmathdigits5000" \
        --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": "juwon1105/RLCR-phi4mini-bigmathdigits5000",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use juwon1105/RLCR-phi4mini-bigmathdigits5000 with Docker Model Runner:
```
docker model run hf.co/juwon1105/RLCR-phi4mini-bigmathdigits5000
```

RLCR-phi4mini-bigmathdigits5000 / README.md

juwon1105

Upload README.md with huggingface_hub

e51f417 verified 2 months ago

preview code

Raw History Blame Contribute Delete

2.18 kB

---
base_model: microsoft/Phi-4-mini-instruct
datasets:
- mehuldamani/big-math-digits
library_name: transformers
tags:
- rl-verify
- grpo
- reasoning
- calibration
license: other
---
# RLCR-phi4mini-bigmathdigits5000

This model is **Phi-4-mini-instruct** trained with **RLCR** (Reinforcement Learning with Calibration Rewards, Damani et al. 2026) on Big-Math-digits: adds a Brier-score-based reward on verbalized confidence to the correctness reward, jointly optimizing reasoning and calibration.

## Results

| Metric | Value |
|---|---|
| Accuracy (held-out) | 0.414 |
| ECE | 0.439 |
| PCE | 0.439 |
| Brier score | 0.427 |
| AUROC | 0.652 |

_Table 4 of the paper. Confidence source for the Phi-4-mini RLCC curricula._

## LoRA

LoRA target modules: `qkv_proj`, `o_proj`, `gate_up_proj`, `down_proj` (explicitly specified).

Rank 16, alpha 32, dropout 0.05.


## Training details

- **Algorithm**: GRPO, 1 epoch, per-device batch size 1, KL penalty β = 0, AdamW (8-bit), max grad norm 1.0.
- **Dataset**: [mehuldamani/big-math-digits](https://huggingface.co/datasets/mehuldamani/big-math-digits), 5,000 training samples, 1,000 held-out for evaluation.
- **Rollouts**: vLLM, temperature 0.7, 32 generations per prompt, effective rollout batch size 1,024.
- **Learning rate**: 5e-6, linear schedule, warmup ratio 0.20.
- **Max response length**: 1,024 tokens.
- **Answer judging**: rule-based (exact match on extracted final answer).
- **Compute**: single RTX 3090 (16-36 GPU-hours) except Qwen3-4B (single A100, 13-19 GPU-hours).


## Citation

This checkpoint accompanies the paper *"Confidence as Curriculum: Reinforcement Learning for Joint Reasoning and Calibration"* (currently under anonymous peer review). A formal citation will be added once the paper is public. It builds directly on:

```bibtex
@inproceedings{damani2026rlcr,
  title     = {Beyond Binary Rewards: Training {LM}s to Reason about their Uncertainty},
  author    = {Damani, Mehul and Puri, Isha and Slocum, Stewart and Shenfeld, Idan and Choshen, Leshem and Kim, Yoon and Andreas, Jacob},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  year      = {2026}
}
```

	---
	base_model: microsoft/Phi-4-mini-instruct
	datasets:
	- mehuldamani/big-math-digits
	library_name: transformers
	tags:
	- rl-verify
	- grpo
	- reasoning
	- calibration
	license: other
	---

	# RLCR-phi4mini-bigmathdigits5000

	This model is Phi-4-mini-instruct trained with RLCR (Reinforcement Learning with Calibration Rewards, Damani et al. 2026) on Big-Math-digits: adds a Brier-score-based reward on verbalized confidence to the correctness reward, jointly optimizing reasoning and calibration.

	## Results

	\| Metric \| Value \|
	\|---\|---\|
	\| Accuracy (held-out) \| 0.414 \|
	\| ECE \| 0.439 \|
	\| PCE \| 0.439 \|
	\| Brier score \| 0.427 \|
	\| AUROC \| 0.652 \|

	_Table 4 of the paper. Confidence source for the Phi-4-mini RLCC curricula._

	## LoRA

	LoRA target modules: `qkv_proj`, `o_proj`, `gate_up_proj`, `down_proj` (explicitly specified).

	Rank 16, alpha 32, dropout 0.05.


	## Training details

	- Algorithm: GRPO, 1 epoch, per-device batch size 1, KL penalty β = 0, AdamW (8-bit), max grad norm 1.0.
	- Dataset: [mehuldamani/big-math-digits](https://huggingface.co/datasets/mehuldamani/big-math-digits), 5,000 training samples, 1,000 held-out for evaluation.
	- Rollouts: vLLM, temperature 0.7, 32 generations per prompt, effective rollout batch size 1,024.
	- Learning rate: 5e-6, linear schedule, warmup ratio 0.20.
	- Max response length: 1,024 tokens.
	- Answer judging: rule-based (exact match on extracted final answer).
	- Compute: single RTX 3090 (16-36 GPU-hours) except Qwen3-4B (single A100, 13-19 GPU-hours).


	## Citation

	This checkpoint accompanies the paper "Confidence as Curriculum: Reinforcement Learning for Joint Reasoning and Calibration" (currently under anonymous peer review). A formal citation will be added once the paper is public. It builds directly on:

	```bibtex
	@inproceedings{damani2026rlcr,
	title = {Beyond Binary Rewards: Training {LM}s to Reason about their Uncertainty},
	author = {Damani, Mehul and Puri, Isha and Slocum, Stewart and Shenfeld, Idan and Choshen, Leshem and Kim, Yoon and Andreas, Jacob},
	booktitle = {The Fourteenth International Conference on Learning Representations},
	year = {2026}
	}
	```